1
Symposium on Proceedings&Abstracts Book of the International Data Science & Engineering
(IDSES’19)
Editör / Editor
Prof. Dr. Filiz ERSÖZ
This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned. Nothing from this publication may be translated, reproduced, stored in a computerized
system or published in any form or in any manner.
Copyright © 2019
Karabük University Publishing, 41
ISBN: 978-605-9554-38-1
http://www.idses.org/ info@idses.org
The individual contributions in this publication and any liabilities arising from them remain the
responsibility of the authors. The publisher is not responsible for possible damages, which could be a
result of content derived from this publication.
i
PREFACE
Dear Colleagues,
On behalf of the Local Organizing Committee I am pleased to welcome our distinguished delegates and
guests to the IDSES’19 – 1st International Data Science and Engineering Symposium (IDSES’19) held
during May 2-3, 2019 in Karabük, TURKEY. IDSES’19 is organized and sponsored by Karabük
University.
The Symposium provides the ideal opportunity to bring together academicians who work in this field,
data scientists, data miners, data engineers and researchers who want to improve themselves. The main
goal of this event is to provide an international scientific forum for the exchange of new ideas in data
science fields. With the increase of global competition and the development of technology, the training
of experts in this field gained importance with the studies carried out in the field of data science and
engineering.
Data discipline and engineering discipline have emerged to give meaning to data stacks, to analyze data
stacks and to transform them into information. The implementation of data science and engineering
methods enables administrators to make effective and quick decisions to increase operational efficiency
as well as to keep the pulse of the society, employees and institutions.
I would like to thank the program committee members for their support at shaping the Symposium
program and the research community for their valuable contributions to the Symposium.
Thank you very much for participating in IDSES’19.
With my warmest regards and respect,
Prof.Dr. Filiz ERSÖZ
Chair of IDSES’19
On behalf of Organizing Committee
May 02-03, 2019 Karabük-TURKEY
ii
Sempozyum Onursal Başkanı / Honorary Head of Symposium
Prof. Dr. Refik Polat, Rector of Karabük University
Sempozyum Başkanı / Chair
Prof. Dr. Filiz ERSÖZ, Director, Institute of Natural and Applied Sciences, Karabük University
Düzenleme Kurulu / Organizing Committee
Prof. Dr. Ahmet OZTÜRK (Uludağ University)
Prof. Dr. Erkan IŞIĞIÇOK (Uludağ University)
Prof. Dr. Mehmet KABAK (Gazi University)
Prof.Dr. Necmi GÜRSAKAL (Fenerbahçe University)
Assoc. Prof. Dr. Canan HAMURKAROĞLU (Karabük University)
Asst. Prof. Dr. Nadi Serhan AYDIN (İstinye University)
Asst. Prof. Dr. Taner ERSÖZ (Karabük University)
Asst. Prof. Dr. Turgut ÖZSEVEN (Tokat Gaziosmanpaşa University)
Bilim Kurulu /Hakemler/ Scientific Committee
Prof.Dr. Ahmet ÖZTÜRK (Uludağ University)
Prof. Dr. Ali GÜNGÖR (Karabuk University)
Prof. Dr. Ali KÖSE (Marmara University)
Prof. Dr. Ali OZDEMIR (Dokuz Eylül University)
Prof. Dr. Ayşe NALLI (Karabük University)
Prof. Dr. Bülent ÖZ (Osmaniye University)
Prof. Dr. Cemalettin KUBAT (Sakarya University)
Prof. Dr. Cevriye GENCER (Gazi University)
iii
Prof. Dr. Dilek ÖNKAL (Bilkent University)
Prof. Dr. Elisabeth PEREIRA (Aveiro University)
Prof. Dr. Emel Kızılkaya AYDOĞAN (Erciyes University)
Prof. Dr. Erdal KILIÇ (Ondokuz Mayıs University)
Prof. Dr. Erkan IŞIĞIÇOK (Uludağ University)
Prof. Dr. Filiz ERSÖZ (Karabük University)
Prof. Dr. Gerhard Wilhelm WEBER (Poznan University)
Prof. Dr. Harun TAŞKIN (Sakarya University)
Prof. Dr. Ineta GEIPELE (Riga University)
Prof. Dr. İhsan YÜKSEL (Kırıkkale University)
Prof. Dr. İsmail Ragıp KARAŞ (Karabük University)
Prof. Dr. José António FILIPE (Lisboa University)
Prof. Dr. Kerim ÇETİNKAYA (Karabük University)
Prof. Dr. Massimo CANALICCHIO (CIA Agricoltori Italiani Umbria)
Prof. Dr. Mehmet KABAK (Gazi University)
Prof. Dr. Mehmet ÖZALP (Karabük University)
Prof. Dr. Metin DAĞDEVİREN (Gazi University)
Prof. Dr. Mücahit COŞKUN (Karabük University)
Prof. Dr. Necmi GÜRSAKAL (Fenerbahçe University)
Prof. Dr. Patrick De CAUSMAECKER (Ke Leuven University)
Prof. Dr. Raif BAYIR (Karabük University)
Prof. Dr. Serpil CULA (Başkent University)
Prof. Dr. Stan URYASEV (University of Florida)
Prof. Dr. Süleyman DÜNDAR (Karabük University)
Prof. Dr. Şenol ALTAN (Gazi University)
Prof. Dr. Şeref SAGIROĞLU (Gazi University)
Prof. Dr. Tatjana TAMBOVCEVA (Riga Technical University)
Prof. Dr. Tülay İlhan HAS (Karadeniz Technical University)
iv
Prof. Dr. Veysi İŞLER (Hasan Kalyoncu University)
Assoc. Prof. Dr. Aslıhan TÜFEKCİ (Gazi University)
Assoc. Prof. Dr. Canan HAMURKAROĞLU (Karabük University)
Assoc. Prof. Dr. Kürşat İŞLEYEN (Gazi University)
Assoc. Prof. Dr. Laura Lotero VELEZ (Universidad Pontificia Bolivariana)
Assoc. Prof. Dr. Ömür TOSUN (Akdeniz University)
Assoc. Prof. Dr. Sadia SAMAR ALI (King Abdul-Aziz University)
Assoc. Prof. Dr. Safiye TURGAY (Sakarya University)
Assoc. Prof. Dr. Semra BORAN (Sakarya University)
Assoc. Prof. Dr. Sezgin IRMAK (Akdeniz University)
Assoc. Prof. Dr. Tolga GENÇ (Marmara University)
Assoc. Prof. Dr. Ufuk COŞKUN (Karabük University)
Assoc. Prof. Dr. Yusuf Tansel İÇ (Başkent University)
Asst. Prof. Dr. Ali Osman ÇIBIKDİKEN (Necmettin Erbakan University)
Asst. Prof. Dr. Cumhur GÜNGÖROĞLU (Karabük University)
Asst. Prof. Dr. Çağrı KOÇ (University of Social Sciences)
Asst. Prof. Dr. Emin Taner ELMAS (İskenderun Technical University)
Asst. Prof. Dr. Hande KÜÇÜKÖNDER (Bartın University)
Asst. Prof. Dr. Linda KAUŠKALE (Riga University)
Asst. Prof. Dr. Murat ALAN (Karabük University)
Asst. Prof. Dr. Mükerrem Bahar BAŞKIR (Bartın University)
Asst. Prof. Dr. Nadi Serhan AYDIN (İstinye University)
Asst. Prof. Dr. Nilesh WARE (Pune University)
Asst. Prof. Dr. Taner ERSÖZ (Karabük University)
Asst. Prof. Dr. Turgut ÖZSEVEN (Tokat Gaziosmanpaşa University)
v
Contents
Factor Analysis of Distribution Tails: Applications in Finance ............................................................... 1
Identification of Green Supply Chain Management (GSCM) Barriers in the Indian Context ................ 4
How Should Data Science Education Be? .............................................................................................. 5
Data Analysis and Kansei Engineering ................................................................................................... 8
Methodology for Building A Security System for Banking Information Resources............................. 11
LiBerated Social Entrepreneur. Using Business Metrics: Migport Refugee Big Data Analytics.
With a Note on Ability and Disability .................................................................................................. 13
Forecasting and Technical Comparison of Inflation in Turkey with Box-Jenkins (ARIMA)
Models and Artificial Neural Networks ............................................................................................... 16
The Importance of Data Mining for Businesses .................................................................................. 17
Factors Affecting the Adoption of Social Networks for Academic Purpose in
Jordanian Universities ......................................................................................................................... 23
An International Framework for a More Sustainable Agriculture: Digital Farming,
Transfer of Innovative Knowledge, Training and Certification of Performances ................................ 25
Facebook Games Applications ............................................................................................................ 33
A Note on the Examination of Portugal’s Hotels Performance - a Proposal for a New
Perspective’s Approach ....................................................................................................................... 44
A Comparative Case Study on Time Series Prediction ........................................................................ 53
Optimal PID-like Fuzzy Logic Controller Design for Ball and Beam System ........................................ 56
Adopting Machine Learning Algorithms for Cloud-Based Application Categorization ....................... 57
A New Nonparametric Test For Testing Equality of Locations Against Umbrella Alternatives........... 60
Smart Agriculture Applications with IoT ............................................................................................. 64
Teknoloji Kabul Modeli Kullanarak Netflix Platformu Kullanma Maksadının Belirleyicileri ............... 72
Meta-Heuristic Methods Used in Optimization of SVM Learning Parameters ................................... 75
Determination Similarities of Basic Financial Indicators of Enterprises Included in the NASDAQ
Index Using by Hierarchical Clustering Distance Methods.................................................................. 87
RFM Model for Segmentation in Retail Analytics: A Case Study......................................................... 90
Siber Tehdit İstihbaratı Alanında Makine Öğrenmesi Algoritmalarının Kullanılması .......................... 93
Analysis of Non-Risked Provinces; Unemployment and Traffic Accidents.......................................... 96
Use of Grid Search in Hyper-Parameter Selection for Time Series Analysis: A Case Study
with Ad Mediation Software ............................................................................................................... 97
1
Marketing and Data Analytics; Increasing Importance of Marketing ............................................... 107
Data Analytics and Importance in Health Sector .............................................................................. 111
Program Development for Cost Calculation in Different Hole Drilling Operations .......................... 124
Multi-Objective Optimization of Hard Turning: Non-Dominated Sorting Genetic Algorithm-II
Approach ........................................................................................................................................... 126
CO2 Emission and Energy Consumption for Different Climate and Building Materials .................... 137
Bitcoin Price Forecasting with Multivariate Long Short Term Memory (LSTM)
Deep Learning Method...................................................................................................................... 140
Covering-Based Generalized IF-Rough Set Models For A Selecting HVAC System ........................... 141
Determination of Production Defects in Iron and Steel Sector by Data Mining ............................... 143
Connected Employee Platform and A Case Study in A Global Company .......................................... 146
Investigation of the Effects of Normal Distribution or Nonnormal of Data on
Machine Capability Analysis .............................................................................................................. 150
Portfolio Selection with the Possibilistic Mean – Variance Model: An Application on the
Borsa Istanbul .................................................................................................................................... 152
Building Digital Assistant (ChatBot) with SAP Conversational Artificial Intelligence ........................ 153
Map Ranking, Map-Reduce and Application in Big Data Analysis..................................................... 163
Benchmarking of OECD Countries in Views of Value-Added Manufacturing Using DEA .................. 165
Recent Literature of Benchmarking of Countries .............................................................................. 166
Benchmarking Analysis of OECD Countries ....................................................................................... 169
A Study on Method Prediction for a Better Directed Treatment of Warts ....................................... 175
Estimation of the Demand for the Blood Bank Using Hybrid PCA-ANFIS Method............................ 187
Analysing of Multivariate Processes with Machine Learning Algorithms ......................................... 189
Comparison of Two Different Social Groups on Twitter with Network Analysis .............................. 191
Analysis of Earthquake Awareness in Education By Data Mining ..................................................... 209
Tersanelerde Yalın 6 Sigma ve Uygulanabilirliği ................................................................................ 213
Smart and Green Supply Chain Applications in Enterprises.............................................................. 215
Bitcoin Price Prediction by Using Artificial Neural Networks and Time Series ................................. 219
Analysis of Presence of Bank Branches According to Settlement in Turkey with Data Mining ....... 227
Artificial Bee Colony Algorithm for Container Loading Problem....................................................... 229
Image Size Scaling and Feature Transformation Function Application for Image Processing
in Machine Learning .......................................................................................................................... 231
Determination of Weights With Fuzzy AHP in the Job Evaluation Process....................................... 232
2
Cost Estimation in the Iron and Steel Industry.................................................................................. 234
Determination of Socio-Economic Factors Affecting Forest Fires (A Case Study of Forest
Regional Directorate of Antalya) ....................................................................................................... 237
Group Acceptance Sampling Plans Based on Time Truncated Life Tests For Compound
Weibull-Exponential Distribution ...................................................................................................... 259
Stochastic Approach to a Buffer Stock Problem ............................................................................... 260
Prediction of Air Permeability of Denim Fabrics Using Articifial Neural Networks........................... 262
Numerical Investigation of Cutting Forces in Turning of C23000 Brass Alloy ................................... 263
Evaluation of Critical Factors in Industry 4.0 Transition Processes by R’WOT Analysis .................... 272
Determination of Variables That Affect the Satisfaction Levels of Visiting Tourists By
Logistics Regression Analysis ............................................................................................................. 274
Demand Estimation of Wood Quantity Used in Wood Industries .................................................... 288
Deep Learning Based Abnormality Detection Application in Enterprise Network Traffic ................ 300
Lean Production in Iron and Steel Production Line ........................................................................... 303
A Road Map for The Big Data Implementation in the Judicial System.............................................. 305
Supplier Selection Using an Intuitionistic Fuzzy Evaluation System ................................................. 306
Environmental Impact Assessment of Two Alternative Wastewater Neutralization Chemicals
in Textile Industry Wastewater Treatment Plant .............................................................................. 307
Empati Çıkar Yöntemiyle Akademik Çalışma Gruplarının İncelenmesi ............................................. 310
Application and Comparison of Biclustering Methods in Detecting Crime Regions ......................... 311
Kronik Böbrek Hastalığının Makine Öğrenmesi Teknikleri ile Sınıflandırılması ................................. 313
Survey on Dynamic Bayesian Network Software Tools..................................................................... 316
Determination of Factors Affecting Employee Productivity ............................................................. 318
3
Factor Analysis of Distribution Tails: Applications in Finance
Stan URYASEV
Director of the RMFE Lab in Industrial & Systems Engineering
University of Florida, USA
Abstract: A popular risk measure, Conditional Value-at-Risk (CVaR), is called Expected
Shortfall (ES) in financial applications. The paper developed algorithms for implementation of
linear regression for estimating CVaR as a function of some factors. Such regression is called
CVaR (Superquantile) regression. The main statement of the paper: CVaR linear regression can
be reduced to minimizing the Rockafellar Error function with linear programming. The theoretical
basis for the analysis is established with the Quadrangle Theory of risk functions. We derived
relationships between elements of CVaR Quadrangle and Mixed-Quantile Quadrangle for discrete
distributions with equally probable atoms. The Deviation in CVaR Quadrangle is an integral. We
presented two equivalent variants of discretization of this integral, which resulted in two sets of
parameters for the Mixed-Quantile Quadrangle. For the first set of parameters, the minimization
of Error from CVaR Quadrangle is equivalent to the minimization of Rockafellar Error from the
Mixed-Quantile Quadrangle. Alternatively, a two-stage procedure based on Decomposition
Theorem can be used for CVaR linear regression with both sets of parameters. This procedure is
valid because the Deviation in the Mixed-Quantile Quadrangle (called Mixed CVaR Deviation)
coincides with the Deviation in CVaR Quadrangle for the both sets of parameters. We illustrated
theoretical results with a case study demonstrating the numerical efficiency of the suggested
approach. The case study codes, data and results are posted at the website. The case study was
done with the Portfolio Safeguard (PSG) optimization package which has precoded Risk,
Deviation, and Error functions for the considered Quadrangles.
Keywords: Quantile, VaR, Quadrangle, CVaR, Conditional Value-at-Risk, Expected Shortfall,
ES, Superquantile, Deviation, Risk, Error, Regret, Minimization, CVaR Estimation, Regression,
Linear Regression, Linear Programming, Portfolio Safeguard, PSG
1
REFERENCES
[1] Acerbi, C. and D. Tasche (2002): On the Coherence of Expected Shortfall, Journal of Banking
and Finance 26, 1487–1503.
[2] Adrian, T., and Brunnermeier, M. (2016): CoVaR. American Economic Review 106 (7), 1705–
1741.
[3] Basset G.W. and H-L Chen (2001): Portfolio Style: Return-based Attribution Using Quantile
Regression. Empirical Economics 26, 293-305.
[4] Carhart M.M. (1997): On Persistence in Mutual Fund Performance. Journal of Finance 52, 5782.
[5] Case Study (2014): Style Classification with Quantile Regression. http://www.ise.ufl.
edu/uryasev/research/testproblems/financial_engineering/style-classification-with-quantile-regression/
[6] Case Study (2016): Estimation of CVaR through Explanatory Factors with CVaR
(Superquantile) Regression.h ttp://www.ise.ufl.edu/uryasev/ research/testproblems/ financial_engineering/
on-implementation-of-cvar-regression/
[6] Huang W.Q. and S. Uryasev (2018). The CoCVaR Approach: Systemic Risk Contribution
Measurement. Journal of Risk. V.20(4), April 2018, DOI:10.21314/JOR.2018.383, 75-93.
[7] Koenker, R. and G. Bassett (1978): Regression Quantiles, Econometrica 46, 33–50.
[8] Koenker, R. (2005): Quantile Regression. Cambridge University Press.
[9] Portfolio Safeguard (2018): http://www.aorda.com
[10] Rockafellar, R.T. and S. Uryasev (2000): Optimization of Conditional Value-At-Risk. The
Journal of Risk, 2(3), 21-41.
[11] Rockafellar, R. T. and S. Uryasev (2002): Conditional Value-at-Risk for General Loss
Distributions, Journal of Banking and Finance 26, 1443-1471.
[12] Rockafellar, R. T., Uryasev, S., and M. Zabarankin (2008): Risk Tuning with Generalized
Linear Regression. Mathematics of Operations Research, 33(3), 712–729.
[13] Rockafellar, R. T., and S. Uryasev (2013): The Fundamental Risk Quadrangle in Risk
Management, Optimization and Statistical Estimation. Surveys in Operations Research and Management
Science, 18, 33–53.
2
[14] Rockafellar, R.T., Royset, J.O., and S.I. Miranda (2014): Superquantile Regression with
Applications to Buffered Reliability, Uncertainty Quantification and Conditional Value-at-Risk. European
J. Operations Research, 234, 140-154.
[15] Rockafellar, R.T., and J.O. Royset (2018): Superquantile/CVaR Risk Measures: Second-order
Theory. Annals of Operations Research, 262, 3-29.
[16] Sharpe W.F. (1992): Asset Allocation: Management Style and Performance Measurement.
Journal of Portfolio Management (Winter), 7-19.
3
Identification of Green Supply Chain Management (GSCM) Barriers in the Indian Context
Mohd. Azmi Khan1, Salma Ahmed2, Sadia Samar Ali3
1,2
Aligarh Muslim University, Department of Business Administration Aligarh,
India
3
King Abdul-Aziz University, Department of Industrial Engineering
Jeddah, Saudi Arabia
Abstract: Due to the new trends in global warming, environmental consciousness has become a
greater concern among the organizations and industries globally. Green Supply Chain
Management (GSCM) has received an increased attention from academia and industries in last
few years. Green Supply Chain Management (GSCM) has emerged as an innovative
organizational strategy to reduce environmental impacts of supply chain activities as well as
efficient usage of energy and material. In today’s developing economies, customers are becoming
more conscious of the environment and governments are making stringent environmental laws, so
the industries need to reduce the environmental impact of their supply chain activities. There are
many barriers which directly and indirectly affect the implementation of GSCM in an organization.
In this paper twenty barriers have been identified through extensive literature review and expert
opinion of academicians. These barriers are found to exist in all organizations irrespective of
industry domain. Due to the presence of various barriers, Indian organizations are struggling to
implement GSCM in their operations. By removing the barriers, Indian industries can focus on
cleaner production by adopting Green Supply Chain Management (GSCM) in their operations.
The Objective of the study was to identify the GSCM barriers in Indian context. The research
methodology was perusing literature in GSCM and validating by experts opinion. Literature was
perused irrespective of industry domain. The study concludes by narrowing on twenty barriers
which play a prominent role in the Indian context.
Keywords: Green Supply Chain Management, Barriers, Indian Context.
4
How Should Data Science Education Be?
Necmi GÜRSAKAL1, Fırat Melih YILMAZ2, Ecem ÖZKAN2, Deniz OKTAY2
1
Fenerbahçe University, Istanbul, Turkey
2
Uludağ University, Bursa, Turkey
Abstract: The interest in data science is increasing in recent years. Data science, including
mathematics, statistics, big data, machine learning, and deep learning; can be considered as the
intersection of statistics, mathematics, and computer science. Although the debate continues about
the core area of data science, the subject is a huge hit. Universities have a high demand for data
science. They are trying to live up to this demand by opening postgraduate and doctoral programs.
Since the subject is a new field, there are significant differences between the programs given by
universities in data science. Besides, since the subject is close to statistics, most of the time, data
science programs are opened in the statistics departments, and this also causes differences between
the programs.
Data science education has to be more project-based since up to now, there is no core knowledge
of data science like other sciences. It is probably the hypercorrect choice to learn this job in a
university which is intertwined with industry and provides plenty of opportunity for internships.
In this article, we will summarize the data science education developments and give curriculum
examples from the world at the undergraduate and graduate level. Regarding these examples, every
university thinks data science as he wants and the names and the contents of these programs really
differs.
Keywords: Data Science, Data Product, Recommendation System.
REFERENCES
[1] S. Gutierrez, Data Scientists at Work, Apress, 2014.
[2] N. Casey, “Can AI fix education? We asked Bill Gates”, 2016.[Online]. Available:
https://www.theverge.com/2016/4/25/11492102/bill-gates-interview-education-software-artificialintelligence [Accessed: 13- April- 2019]
5
[3] Atlantic, “Artificial Intelligence Promises a Personalized Education forAll”,2018.[Online].
Available:https://www.theatlantic.com/sponsored/vmware-2017/personalized-education/1667/ [Accessed:
13- April-2019]
[4] S. Lior, "Is Data Science an Academic Discipline?”,2017.[Online]. Available: https://www.
datasciencecentral.com /profiles /blogs/is-datascience-an-academic-discipline [Accessed: 13- April- 2019].
[5] P. L. Jennifer, “Maslow’s Hierarchy of Data Science: Why Math and Science Still Matter”,
2019, [Online]. Available: https://www.datasciencecentral.com/profiles/blogs/maslow-s-hierarchy-ofdata-science-why-math-and-science-still [Accessed: 13- April- 2019].
[6] E. Colson, “Why Data Science Teams Need Generalists”, HBR, 2019, [Online]. Available:
https://hbr.org/2019/03/why-data-scienceteams-need-generalists-not-specialists [Accessed: 13- April2019].
[7] N. Thomas, “How It Feels to Learn Data Science in 2019”, Towards Data Science, 2019.
[Online].
Available:
https://towardsdatascience.com/how-it-feels-to-learn-data-science-in-2019-
6ee688498029. [Accessed: 15- April- 2019].
[8] DJ Patil, Building Data Science Teams, O’Reilly Media, 2011.
[9] S. Steve, “Enable Deeper Understanding with Great Data Storytelling”,2017.[Online].
Available: https://tdwi.org/articles/2017/03/09/enable-deeper-understanding-with-great-data-storytelling.
aspx [Accessed: 15- April- 2019].
[10] S.C. Hicks, and R. A. Irizarry, “A Guide to Teaching Data Science.”, American Statistical
Association, 2018. Pp.382–91.
[11] A. Stone, “Will Data Scientists Have a Big Impact on Education?”, 2017. Online]. Available:
https://www.govtech.com/education/k-12/Will-Data-Scientists-Have-a-Big-Impact-on-Education.html.
[Accessed: 15- April- 2019].
[12] W. Vorhies, “Getting a Data Science Education”, Data Science Central, 2015.[Online].
Available:https://www.datasciencecentral.com/ profiles/ blogs/getting-a-datascience-education [Accessed:
17- April- 2019].
[13] A. Woodie, “Universities Get Creative with Data Science Education”, Datanami,
2018.[Online]. Available: https://www.datanami.com/2018/08/23/universities-get-creative- withdatascience-education/ [Accessed: 19- April- 2019].
6
[14] National Academies of Sciences, Engineering, and Medicine. Data Science for
Undergraduates: Opportunities and Options, The National Academies Press. 2018.
[15] Park City Math Institute (PCMI) Undergraduate Faculty Group, “Curriculum Guidelines for
Undergraduate Programs in Data Science”, Annual Review of Statistics, 2016. pp.1–26
[16] T. Akimichi, “A new era of statistics and data science education in Japanese universities”,
Japanese Journal of Statistics and Data Science, 2018, pp. 109–116.
[17] Zhang Jilong, Fu Anna, Wang Hao, Yin Shenqing, “The Development of Data Science
Education in China from the LIS Perspective”, The International Journal of Librarianship,2017, pp. 19–26.
[18] MNM, “20 Universities for pursuing Master of Science in Data Science (On-Campus) in the
USA — 2018”, Towards Data Science,2018. Online]. Available: https://towardsdatascience.com/20universities-for-pursuing-master-of-science-in-data-science-oncampus-in-the-usa-2018-9970d5d25bd5
[Accessed: 22- April- 2019]
[19] University of Waterloo, “Department of Statistics and Actuarial Science”, 2017-2018.
[Online]. Available: https://uwaterloo.ca/statistics-and-actuarial-science/sites/ca.statisticsand-actuarialscience/files/uploads/files/datascience-2017-2018.pdf [Accessed: 22- April- 2019]
[20] Data Science Degree Programs, “30 Best Master’s In Data Scıence Degree Programs 2019”.
2019. [Online]. Available: https://www.datasciencedegreeprograms.net/rankings/masters-datascience/
[Accessed: 22- April- 2019]
[21] J. Brooks, “Why so many data scientists leaving their jobs”, 2018. [Online]. Available:
https://www.kdnuggets.com/2018/04/why-datascientists-leaving-jobs.html [Accessed: 22- April- 2019]
7
Data Analysis and Kansei Engineering
Mustafa Umut ÖZTÜRK, Ahmet ÖZTÜRK
Department of Econometrics, Uludağ University, TURKEY
Abstract: One of the most important steps in establishing a successful business is to do accurate
data analysis. The correct analysis of the data and the correct information as a result of the analysis
reveal the wishes, feelings, emotions, and demands of the users. With data analysis useful
information should be discovered, those who are useless should be cleaned and what should have
done in the next process. Several types of data analysis methods can be done based on the data
collected from Kansei survey. These analyses play an important role in the process of Kansei
Engineering. There are several types of statistical analysis that are developed to use in Kansei
studies such as variance analysis, linear regression analysis, flow data analysis, principal
component analysis, quantification theory I, factor analysis, cluster analysis, rough set theory.
The purpose of data analysis is to synthesis statistical data or Kansei words with the product
properties and therefore to be applied in the design context.
Kansei engineering is a method used to convert a customer’s ambiguous imagine product
into detail design. Kansei Engineering starts from decision of strategy as design domain as well
as target. It is collected the Kansei words related to the product domain. The word Kansei, which
is used in design and other research areas. It means the feeling of beauty and pleasant emotions
reflected by any object and its desire in Japanese. Kansei words form the basis of Kansei
engineering. In a way, Kansei engineering is a product development method which can measure a
customer’s feeling, values and affective needs and translate them into concrete product solutions.
Since 1980’s Kansei Engineering has expanded greatly and become a significant discipline
both in the industrial and the academic world. Furthermore, Kansei Engineering developed as an
efficient research discipline, providing many innovations and market success in the industrial
world. It is founded by Mitsou Nagamachi, a professor at Hiroshima University. He is considered
to be the father of Kansei Engineering. The term Kansei Engineering itself was used the first time
in 1986 by Yamanota, president of Mazda Automotive Corporation at Michigan University.
8
The main aim of this study is to explain Kansei engineering and model which is rarely seen
in Turkish literature and to reveal its relationship with data analysis. Besides, the future of
importance of data, Big data, data analysis and Kansei Engineering will be discussed.
Keywords: Data, Data Analysis, Big Data, Kansei, Chise, Kansei Engineering.
REFERENCES
[1] Data Analysis – Wikipedia, https//en. Wikipedia. Org/wiki/Data analysis.
[2] Doğan, K.- Arslantekin, S., Büyük Veri: Önemi, Yapısı ve Günümüzdeki Durum (Big Data: Its
Importance, Structure and Current Status), Dil Tarih Coğrafya Dergisi, 56.1, 2016, s.15.
[3] Erlag, H.- Kör, H., Büyük Veri ve Büyük Verinin Analizi (Big Data and Its Analysis),
International Conference on Science and Technology, October 3-6, 2016, Ankara.
[4] Harari, Yuval Noah, 21 Yüzyıl için 21 Ders, (Türkçesi, Selin Seral, 21 Lessons fort he 21st
Century) Kollektif Kitap, İstanbul,2018, s.84.
[5] Le’vy, Pierre,” Beyond Kansei Engineering: The Emancipation of Kansei Design”, International
Journal of Design Vol.7, No:2, 2013, p.83.
[6] Lokman, Anitawati Mohd- Nagamachi, Mitsuo,” Validation of Kansei Engineering: Adoption in
E- Commerce Web Design”, Kansei Engineering Internatonal Vol.9 No:1,2009, pp.24.
[7] Lokman, Anitawati Mohd,” Desing and Emotion: The Kansei Engineering Methodology”,
Design: the Kansei Engineering Methodology, Vol.1, Issue1, 2010, pp.1, https:// aniwati.uitm.
edu.my/mypapers/21MJOC10-Designand Emotion-The KE methedology.pdf.
[8] Lokman, Anitawati Mohd,” Design and Emotion: The Kansei Engineering Methedology”, Vol.1,
Issue, 2010, pp.3. https://Aniwati-uitm.edu.my/mypapers/21-MJOC
10 -Design and Emotion- KE
Methodology.Pdf.
[9] Lokman, Anitawati Mohd- Mitsuo Nagamachi,” Validation of Kansei Engineering Adoptatation
in E- commerce Web Design”, Kansei Engineering International Vol.9 No.1, 2009, pp.24.
[10] Marco, Lluis- Tort, Almogro Xavier - Llabre’s, Martorell, “Statistical Methods in Kansei
Engineering: a Case of of Statistical Engineering”, ENBIS 11, September 2011, p.2.
[11] Nagamachi, Mitsuo, Home Applications of Kansei Engineering in Japon: An Overview,
2016:15.4, p.209, https//www. Research gate.net/publication/ 311707844, Home application of Kansei
Engineering in Japon.
9
[12] Okamoto, Ricordo Hirata, Nagamachi, Mitsou, Ishihara, Shigekazu,” Satisfing Emotional Needs
of
the
Beer
consumer
Trough
Kansei
Engineering”,
https://www.research
gate.net//publication/314041244.Satisfying, emotional needs of the beer –consumer through Kansei
Engineering,2004, pp.1-8.,
[13] Öztürk, Ahmet, Kalite Yönetimi ve Planlaması (Quality Management and Planning), 2. Baskı
Ekin Kitabevi, Bursa,2013, p.4.
[14] Seung Hee Lee, Akira Harada-Pieter Jan Stappers, Pleasure with Products: Design based on
Kansei.
https://www.researchgate.net/publication/228396068-Pleasure-with-product-Design-based-on-
Kansei.
[15] Shaari, Nazlina, Methods of Analzing Images based on Kansei Engineering, International
Journal of Computer Science and Electronic Engineering (IJCSEE) Volume 1, Issue3, s.17.
[16] Shafieyoun, Zhabis- Maicocchi, Marco, Flow Kansei Engineering Qualifying conscious and
unconscious behaviour to gain optimal experience in Kansei engineering, International Conference on
Kansei Engineering and Emotion Research, KEER 2014, Linköping, June 11-13, 2014, pp.621.
[17] Schütte, Simon, Designing Feeling into Products Integrating Kansei, Engineering Methodology
in Product Development, Thesis No.946, Institute of Technology, Linköping, 2002, p.23.
[18] Schütte, Simon, Engineering Emotional Values in Product Design: Kansei Engineering in
Development, Linköpin University, Linköpin,2005, pp.45. https://pdfs. Semantic scholar.org/52ao/5
fed.774a1.
[19] Sridhar, Jay, What is the Data Analysis and Why is it important?, Feb.12,2018,https://make
useofcom/tag/what-is-data-analysis/
10
Methodology for Building A Security System for Banking Information Resources
Serhii Yevseiev
Simon Kuznets Kharkiv National University of Economics, UKRAINE
Abstract: The revolutionary changes of the last decade in the banking sector have led to the
integration of information and computer networks into a single information and cybernetic space,
which has led to the creation of automated banking systems that have substantially expanded the
spectrum of electronic services of state and commercial banks of the world. As a result, threats to
such a national information resource of the state as the banking information resources under which
the banking information refers. Threats to the security of banking information resources have
become signs of hybridization. Manifestations of hybridity as a result of the simultaneous impact
of threats to information security, cybernetic security and information security on banking
information resources have led to the emergence of synergies, the negative manifestations of which
require a radical revision of the concepts of the construction of existing security systems. Thus, it
becomes clear that there is a need for a radical revision of the current methodological principles
for building a security system for banking information resources both for Ukraine and for the world
as a whole.
Keywords: Security of Banking Information Resources, Automated Banking System, A
Synergetic Model of Threats to the Security of Banking Information Resources, A Classifier of
Threats to the Security of Banking Information Resources, Information Security, Cybersecurity,
Security to Information, Investments, Emergent Properties, Synergistic Effect, Hybrid Crypto
Codes on Damaged Codes, Elliptic Codes, Methodology.
REFERENCES
[1] Hryshchuk R., Yevseiev, S. Shmatko A. Construction Methodology of Information Security
System of Banking Information in Automated Banking Systems: Monograph, 284 P., Vienna.: Premier
Publishing S. R. O., 2018.
[2] Yudin, O., “Informacion Of Security. Regulatory and Legal Regulations”, 640p., NAU, Kiyev,
2011. 3. Grischuk, R., Danik, Yu., “The Foundations of CyberSec”, ZhNAEU, Zhytomyr, 636 p., 2016.
11
[3] Evseev, S., Korol, O., Kots, G., “Analysis of the Legislative Framework for the NSMEP
Information Security Management System”, East European Journal of Advanced Technologies, v.77, no.
5/3, p. 48–59, 2015.
[4] Evseev, S., “Methodology for Evaluating the Security of Information Technologies of Automated
Banking Systems of Ukraine”, Science and Technology Journal “Information Security”, v. 22, no. 2, p.
297-309, 2016.
[5] Briones, P., Chamoso, A., Barriuso, A., “Review of the Main Security Problems with Multi-Agent
Systems used in E-commerce Applications”, ADCAIJ, Regular Issue, v. 5, no. 3, pр. 55-61, 2016.
[6] Simpson, W., “Securing Information Systems in an Uncertain World Enterprise Level Security”,
Systemics, Cybernetics аnd Informatics, v. 14, no. 2, рр. 83-90, 2016.
[7] Evseev, S., Korol, O., Rzayev, H., Imanova, Z., “Development of a Mac-Alice Modified
Asymmetric Crypto-Code System on Shortened Elliptic Codes”, East European Journal of Advanced
Technologies, v. 4, no. 82, p. 18-26, 2016.
[8] Evseev, S., Korol, O., Kots, G., “Construction of the hybrid systems”, East-European Journal of
Advanced Technologies, v. 4, no. 88, p. 4-21, 2017.
[10] Evseev, S., “The Use of Flawed Codes in Crypto-Code Systems”, Information Processing
Systems, V. 151, no. 5, 2017.
[11] Scheau, C., Arsene, A., Dinca, G., “Phishing and E-Commerce: An Information Security
Management Problem”, Journal of Defence Resources Management, v.7, no. 1 (12), рр. 129-140, 2016.
[12] Alhothaily, Ab., Alrawais, A., Song, T., Lin, B., Cheng, X., “QuickCash: Secure Transfer
Payment Systems”, Sensors, no. 17, рр.1-20, 2017.
[13] Yusupova, O., “Transaction Security When Using Internet Banking”, Financial Analytics:
Problems and Solutions, no. 35, p. 26–40, 2016.
[14] Evseev, S., Kots, G., Lekarev, E., “Developing A Multi-Factor Authentication Method Based on
The Crypto-Code System”, East European Journal of Advanced Technologies, v. 6/4, no. 84, p. 11-23,
2016.
[15] Evseev, S., Korol, O., Kots, G., Minukhin, S., Kholodkova, A., “Eastern European Journal of
Advanced Technologies, v. 5 / 9, no. 89, p. 19–36, 2017.
12
LiBerated Social Entrepreneur. Using Business Metrics: Migport Refugee Big Data
Analytics. With a Note on Ability and Disability
Gerhard-Wilhelm WEBER1, Berat KJAMILI2, Dominik CZERKAWSKI1
Poznan University of Technology1, POLAND; Middle East Technical University2, TURKEY
Abstract: LiBerated Social Entrepreneurship in Developing and Emerging Countries consists of
a social entrepreneur using business metrics, to sustain social impact. We study differences
between developing and developed countries, introducing a new OR approach to development.
Commercial entrepreneurs are generally oriented to business metrics like profit, revenues and
return. Instead, social entrepreneurs are non-profits or a blend with for-profit goals, generating
Return to Society. In DCs, a social entrepreneurship has been uncommon. We introduce a midway as LiBerated Social Entrepreneur, where social businesses should be sustainable. We apply
Game and Max-Flow - Min-Cut Theories, Schumpeter’s creative destruction and Adam Smith’s
diversification model for our business plan. As a result, B. Kjamili started Migport, formerly QZenobia: a mobile application that runs as a “refugee portal”, supported by “Refugee Big-Data
Analytics”: refugees submit data to the application via “questionnaire” and search for
opportunities, verified news privatized based on their answers. The idea of both-sided help with
benefit generated by D. Czerkawski is an extension of B. Kjamili's conception. Nshareplatform
(NSP) will create a friendly public space for people with disabilities, understanding they needs. It
tries to facilitate better communication between “two worlds”- Ability and Disability and
personalizes an assistant (Special person helping people with disabilities). Multivariate Adaptive
Regression Splines (MARS), Conic MARS (CMARS) and its robust version RCMARS have
shown their potential for Big-Data and, recently, Small-Data. With that toolbox, we aim to further
support our joint and novel project.
Keywords: Social Entrepreneur, Start-up, Business Canvas Model, Ability, Disability, OR, Data
mining, Analytics
13
REFERENCES
[1] Ankara Department Agency (2017). TechAnkara Proje Pazarı 2017'de sergilenecek 100 proje
belirlendi. Retrieved from: http://www.ankaraka.org.tr/tr/techankara-projepazari-2017de-sergilenecek100-proje-belirlendi_3810.htm
[2] B. Kjamili and G.-W. Weber, The Role of LiBerated Social Entrepreneur in Developing
Countries: A mid-way, in Societal Complexity, Data Mining and Gaming; State-of-the-Art 2017, Greenhill
& Waterfront, Europe: Amsterdam, The Netherlands; Guilford, UK North-America: Montreal, Canada,
2017. ISBN /EAN 978-90-77171-54-7.
[3] B. Kjamili, G. W. Weber (2014). IFORS: Opening Doors to International Students, Retrieve
from: http://ifors.org/newsletter/ifors-dec-2014.pdf.
[4] Brooks A. C. (2009). Social Entrepreneurship: A Modern Approach to Social Value Creation
(1st ed.). New Jersey, Pearson Education.
[5] Business Dictionary, Inc (2017). Definition of Entrepreneurship Retrieved from
http://www.businessdictionary.com/definition/entrepreneurship.html.
[6] C. Schinckus (2015). The Valuation of Social Impact Bonds: An Introductory Perspective with
Peterborough SIB. Elsevier B.V.
[7] M. Young, The Technical Writer’s Handbook. Mill Valley, CA: University Science, 1989.
[8] Egirisim (2017). Startup Istambul 2017’de İkinci Aşamaya Geçen 50 Girişim. Retrieved from:
https://egirisim.com/2017/10/21/startupistanbul-2017de-ikinci-asamaya-gecen-50-girisim/
[9]
Empowerment
Plan,
Inc
(2017)
Story.
Retrieved
from
http://www.empowermentplan.org/about.
[10] Entrepreneurs’ Organization (2018). Qzenobia To Run For Finals İn MIT Enterprise Forum
Before GSEA| Retrieved from: https://www.eonetwork.org/turkey/chapterpressreleases/qzenobia-torunfor-finals-in-mit-enterprise-forum-before-gsea.
[11] G. Eusden (2013). Max-Flow, Min-Cut: History and Concepts Behind the Max-Flow, MinCut
Theorem
in
Graph
Theory.
Retrieved
Mathematics/sjmiller/public_html/hudson/Eusden_maxflowminut.pdf.
14
from
http://web.williams.edu/
[12] MIT Enterprise Forum (2018). Innovate for Refugees Semifinalists. Retrieved from:
https://innovateforrefugees.mitefarab.org/en/site/semifinalists.
[13] Moblobi (2017). Türkiyedeki Mültecilerin Oluşturduğu Pazar Hakkında Güvenilir Veriler
Sağlayan
Girişim
QZenobia
ile
Röportajımız!
Retrieved
from:
http://moblobi.com/roportajlar/qzenobiaroportaj.html
[14] Nacional Geographic (2006). Nobel Peace Prize Goes to Micro-Loan Pioneers, Retrieved from
http://news.nationalgeographic.com/news/2006/10/061013-nobelpeace.html
[15] ODTU Teknokent (2017). TechAnkara Proje Pazarı 2017’den ODTÜ Teknokent Firmaları
Ödülle Döndü. Retrieved from: http://odtuteknokent.com.tr/tr/haber/techankaraproje-pazari-2017denodtu-teknokent-firmalari-odulle-dondu
[16] Osterwalder A., Pigneur Y. (2010). Business Model Generation: A Handbook for Visionaries
Game Changers, and Challengers, Paperback (1st ed.). New Jersey, John Wiley & Sons.
[17] P. F. Cuevas (2017). The World Bank’s mission: Eradicate Poverty and Boost Shared
Prosperity. Ankara, World Bank group Poverty.
[18]
Strategyzer,
Inc
(2017).
The
Business
Model
Canvas
Website:
https://strategyzer.com/canvas/business-model-canvas
[19] Tech.co, Inc (2015). Top 15 Most Entrepreneurial Countries in the World Website:
http://tech.co/top-15-entrepreneurial-countries-world2015-06
[20] Acumen (2010). An Introduction to Human-Centered Design, Retrieved from
https://www.plusacumen.org/courses/introduction-human-centereddesign
[21]
Liviing
Whith
Disability
(2016).
The
New
York
Times,
Retrieved
from
https://www.nytimes.com/2016/08/28/opinion/living-withdisability.html
[22] Kaul I., Isabelle Grunberg, Marc A. Stern (1999) Global Public Goods: International.
Cooperation in the 21st Century, United Nations Development Program, New York.
[23] Mackelprang, R. and Salsgiver, R. O. (1999) Disability: A Diversity Model Approach in
Human. Service Practice, Brooks/Cole Publishing Company, Toronto.
15
Forecasting and Technical Comparison of Inflation in Turkey with Box-Jenkins (ARIMA)
Models and Artificial Neural Networks
Erkan IŞIĞIÇOK, Ramazan ÖZ, Savaş TARKUN
Department of Econometrics, Uludağ University, TURKEY
Abstract: Inflation refers to an ongoing and overall comprehensive increase in the overall level
of goods and services price in the economy. Today; inflation, which is tried to be kept under control
by the central banks, is trying to ensure price stability, the continuous price changes that arise in
all the goods or services that consumers use includes. Undoubtedly in terms of economy, inflation
expectations are also ganing importance, except for rhe realized inflation. This situation makes it
necessary to predict the future vaules of inflation. In that case, a reliable estimate of the future
values of inflation in any country will create an entry in determining the policies that decisionmaker units will implement on the economy.
The aim of this article is to predict inflation in the next period by using the Consumer Price Index
(CPI) data with two alternative techniques. It is also aimed to examine the prediction performances
of these two techniques in comparisons. Thus, the first of the two main objectives of the study is
to predict the future values of inflation with two alternative techniques. The second goal is to
determine which of these two techniques well compared to statistical and econometric criteria.
In this context, the estimated performance of both techniques was predicted by the 9-month
inflation, Box-Jenkins (ARIMA) and Artificial Neural Networks (ANN) in the April – December
2019 period, using CPI data consisting of 207 in the period of January 2002 – March 2019. In the
study, Eviews and Matlab programs were utilized.
Keywords:
Inflation,
Box-JEnkins, ARIMA, Artificial Neural
(Forecasting), Technical Comparison.
16
Networks,
Prediction
The Importance of Data Mining for Businesses
Filiz ERSÖZ
Department of Industrial Engineering, Karabük University, TURKEY
Abstract: Today, with digitalization, it is possible to read digital data and make the right decisions
based on analytical results. Along with big data, the science of data management and analysis is
evolving to enable organizations to transform their knowledge into information that will help them
achieve their goals. In this study, it is given as an example to increase awareness of big data, data
mining, data mining and its applications in various sectors in Turkey.
Keywords: Industry 4.0, Digitalization, Big Data, Data Mining.
Introduction
The mportance of data has started all over the world and w th the ncrease of b g data every day,
data process ng has become more mportant. The data revolut on, affect ng the ent re world and
all sectors, has attracted attent on w th ncreas ng technology and advances n mach ne learn ng.
In add t on, the ncreas ng number of structured and unstructured large data owned by enterpr ses
n recent years has led to the need to make sense of these data. Systems that support data volumes
along w th b g data cont nued to ncrease rap dly. M crosoft s nvolved n bus ness analyt cs, data
sc ence and mach ne learn ng, b g data systems and platforms or data management.
Big companies are now making great use of technology and digitalization to solve their
problems and make predictions for the future. Using big data and management, these companies
make sense and analyze data based on digital and artificial intelligence and shed light on future
planning. However, SMEs hesitate to adopt data science and these technologies. SMEs do not have
enough resources for artificial intelligence or digital applications and analytics. In addition,
analytical approaches are approached with suspicion.
Researchers have revealed the term "big data" and data mining (Data analytics) to describe
this evolving technology. In other words, it can be defined as a repository of information that
17
provide predictive results that will solve the business problem or develop the strategies of the
business and enable the business to make good decisions. The data size is still Zettabyte size [1]
and this data is processed and guides the enterprises.
Data mining, data analytics, and big data are essentially data science-related sciences.
Today, it is used in many places in the same function and in the same sense. Big data and analytics
have increased rapidly in the service and manufacturing sectors. Data storage technology, data
processing technology, data visualization technique, models and algorithms, and in particular the
creation of the right decision-making models, offers opportunities for service and manufacturing
enterprises [2].
Data science is said to be an industry with a market value of more than $ 9 trillion by 2020,
and not only the ability to extract relevant information, but also the ability to make accurate
estimates and significantly improve strategies and performance in the industry [3]. The fields
associated with data Science (analytics) are given in Figure 1. Data science is said to be an industry
with a market value of more than $ 9 trillion by 2020, and not only the ability to extract relevant
information, but also the ability to make accurate estimates and significantly improve strategies
and performance in the industry [3]. Scientific methods, visualization, statistical modeling and
calculation, data technology, data consultancy.
Firstly, Big data & technologies and analytics are used extensively in many sectors such as
health, management, banking and finance, manufacturing, insurance, electronic commerce,
communication, transportation, defense, fraud detection and education.
Today, organizations are confronted with complex databases due to the development of
technology, the increase of databases and information technologies, and the widespread use of
information technologies. It is clear that if the data in accordance with the needs of the institutions
is managed successfully and effectively, it will offer great advantages and opportunities to
institutions and organizations in economic terms. Each action performed in a digital environment
leaves behind a data record. In fact, with each step taken and every choice made, new data is
created. According to IDC (International Data Corporation) Big Data and Business Analytics
Forum data, the registered data volume increased to 16 ZB (1 ZB 1.09 Trillion Gigabytes) in 2016,
it is estimated that the data record will be 35 Zettabyte in 2020 and 163 Zettabyte in 2025 (1024
ZB = 1 Yottabyte (YB)).
18
In recent years, investment projects for transformational information technologies for
different technologies in the financial sector, energy sector, healthcare sector, telecom sector and
public institutions have started to increase rapidly. These technologies need to be structured in data
centers to meet their corporate goals, increase customer satisfaction and respond to business needs.
Rapid walk in projects such as national Data center or city hospitals in Turkey, our transition to
4.5 G, continued rapid growth of cloud computing, information technologies in the formation of
future data centers, will further increase the importance of big data and data mining (Business
Analytics). In addition, with the increasing complexity of the decision-making process and the
need for more numerical and textual data, it became difficult to reveal valuable and meaningful
information in big data bases. Amounts of data’s being in very large quantity (algorithms for
processing data such as Zettabyte, Petabyte, Terabyte must be highly measurable), high size of
data (tens of thousands can be micro-arrays), data’s being very complex, presence of new and
advanced applications requires data mining and text mining.
The data collected in the data stacks stored in databases and data warehouses is now very
large. The need to uncover meaningful relationships, patterns and trends from big data stacks has
increased the importance of processing data in making accurate and strategic decisions. For these
reasons, in data mining and text mining application studies, the value of "machine learning"
techniques and "statistical analysis and modelling" has also increased in parallel.
Data mining is the process of discovering the rules and patterns associated with each other
from big data stacks. It is not just a technique; it is a data approach that accommodates many
techniques. Converts all information from data stacks to an easy and understandable structure.
Data mining is related to both database techniques and machine learning. Information from data is
the extraction of valuable information in short, data mining can be defined as "the way to convert
data to qualified information".
Today, data mining is also referred to as "business intelligence" and "Business Analytics"
as well. Other definitions are; Knowledge mining from the databases, Knowledge extraction, Data
/ pattern analysis, Data archeology and Data analytics.
Data mining; It is an interdisciplinary study where machine learning, statistics, database
technology, artificial intelligence and visualization are used together. The most important of these
19
fields is “the science of statistics”. Statistical methods are the basis for the data mining tools and
methods that are being used today.
In data mining; with a study aimed at achieving specific results from large and meaningless
masses of data; The data is passed through several stages prior to modeling. The first step is to
clean the data before modeling. With the determination of outliers and extreme values, after clean
and quality data is obtained (Cleaning), combining the data enable to be able to speak the same
language. Here, the relevant and important variables for the research topic are selected and size
reduction is performed. As a result of the analysis, the transformation of the available data into a
format suitable for reuse, evaluation of the importance of data and relations (Evaluation) and
presentation of the results to the decision makers (Presentation) are the processes that complete
the data mining. Stages of data mining; Starting from the database, the transformation of data into
information is given below [4].
Figure 2. Process of Acces to Information
The data mining cycle is completed by withdrawing information from databases and the
results of the analysis with the interpretation of the decision maker. The Data Mining process can
be expressed as a step of the process called “Knowledge Discovery in Databases” and “Decision
Support System”.
20
In data mining, business information must be used together with advanced information
technologies for the disclosure of information within the database.
Conclusion
Data mining contributes to all sectors. Today, data mining, data analytics and data science, together
with concepts such as, but will play an important role in the decision making and roadmaps of
production or service businesses. The benefits of data mining to enterprises are briefly given
below.
Data mining contributes to all sectors. Today, the concept of data mining is associated with
concepts such as data analytics and data science. These concepts will play an important role in the
correct decision making and roadmaps of production or service enterprises in the future.
Data mining in enterprises;
•
It reveals the valuable information in enterprises by understanding the general
computations and probability principles underlying the data in big data stacks, modern
machine learning and mining algorithms.
•
Analyzes data for scientific and business Analytics with the implementation of many
computational and statistical methods.
•
Enables the determination and resolution of the appropriate method for the collection and
use of data in enterprises.
•
By implementing machine learning (software programs for private data mining) solutions
enable enterprises to discover new and efficient information.
•
Helps the decision maker to make a good decision and to report the information in the
enterprises in a clear and understandable way.
•
By using data mining and technologies in enterprises that have a big data bound, they solve
their problems, reveal their needs, plan and develop their strategies for the future.
•
Production or service businesses bring out the need not only to gain customers but also to
develop long-standing relationships by optimizing the experiences of their customers.
•
It provides efficiency and efficiency in meeting customer expectations through analytical
applications in designing, controlling and optimizing business operations in the production
of goods or services.
21
•
It helps businesses quickly identify fraud by improving their data and analytical
capabilities. Provides continuous monitoring of activities based on their prediction and
determination of future activities.
•
With forecasting analysis, it can reduce the business risk or out-of-service risk of
businesses. In particular, retail and service-based businesses can use predictive analytics
to better understand the success of new products or with whom they do business.
REFERENCES
[1] Prof. Dr. Necmi Gürsakal, Büyük Veri. Bursa: Dora Yayınları, 2017.
[2] Ray Y. Zhong, Stephen T. Newman, George Q. Huang, “Big Data for Supply Chain
Management in the Service and Manufacturing Sectors: Challenges, Opportunities, and Future
Perspectives,” Comput. Ind. Eng., vol. 101, pp. 572–591, 2016.
[3] Victor Roman, “4.0 Industry Technologies & Supply Chain” [Online]. Available:
https://towardsdatascience.com/4-0-industry-technologies-supply-chain-97c857de14ae.
[4] Han, J. and Kamber, M. (2006) Data Mining Concepts and Techniques. 2nd Edition, Morgan
Kaufmann Publishers, San Francisco.
22
Factors Affecting the Adoption of Social Networks for Academic Purpose in Jordanian
Universities
Ala’a Abu Gharrah GHARRAH, Ali ALJAAFREH, Noor AL-MA’AITAH
School of Business, Mutah University, JORDAN
Abstract: Due to the rapid revolution in information technology (IT), teaching methods differ
from those used in the past. In recent years, Social Networks (SNs) have become very popular
among people. SNs such as Facebook and Twitter can be used in the learning process to stimulate
thoughtful discussions on specific classroom topics, and to share learning resources. Despite the
Universities have their own eLearning platforms; students are using SNs for the same purpose.
The current study attempts to explore the factors affect students’ adoption to use SNs for academic
purposes.
Keywords: Social Network, Higher Education, Jordan
REFERENCES
[1] Balakrishnan, V. (2017). Key Determinants For İntention to Use Social Media For Learning İn
Higher Education İnstitutions. Universal Access İn The Information Society, 16(2), 289-301.
[2] Tan, M., Shao, P., & Yu, P. (2014). Factors Influencing Engineering Students’ Use of Social
Media In Learning. World Trans. Eng. Technol. Educ, 12(4), 648-654.
[3] Boyd, D. M., & Ellison, N. B. (2007). Social Network Sites: Definition, History, And
Scholarship. Journal Of Computer‐Mediated Communication, 13(1), 210-230.
[4] Kemp, S. (2018). Digital İn 2018: World’s Internet Users Pass The 4 Billion Mark. Retrieved
From Https://Wearesocial.Com/Blog/2018/01/Global-Digital-Report2018
[5] Seely Brown, J., & Adler, R. (2008). Open Education, The Long Tail, And Learning 2.0.
Educause Review, 43(1), 16-20.
[6] Kaplan, A. M., & Haenlein, M. (2012). Social Media: Back to The Roots And Back To The
Future. Journal Of Systems And Information Technology, 14(2), 101-104.
23
[7] lStats, S. C. G. Browser Market Share Worldwide (Mar 2018 – Mar 2019). Available from:
http://gs.statcounter.com/social-mediastats/all/jordan
[8] Selwyn, N. (2009). Faceworking: Exploring Students' Education‐Related Use of Facebook.
Learning, Media And Technology, 34(2), 157-174.
[9] Tapscott, D., & Williams, A. D. (2010). Innovating the 21st-Century University: It’s Time.
Educause Review, 45(1), 16-29.
[10] Malkawi, N. M., & Halasa, A. (2016). Exploiting Electronic Social Networks in Educational
Process: Study at Universities in Irbid State-Jordan.
[11] Hudson, S., & Thal, K. (2013). The Impact of Social Media on The Consumer Decision
Process: Implications For Tourism Marketing. Journal of Travel & Tourism Marketing, 30(1-2), 156-160.
[12] Li, W., & Darban, A. (2012). The Impact of Online Social Networks on Consumers'
Purchasing Decision: The Study Of Food Retailers.
[13] Bonilla Polo, P. A., & Osman, M. (2017). Is Social The New Smart? Factors Influencing
Students on Their Decision to Use Social Media For Academic Purposes.
24
An International Framework for a More Sustainable Agriculture: Digital Farming,
Transfer of Innovative Knowledge, Training and Certification of Performances
Massimo CANALICCHIO
CIA Agricoltori Italiani Umbria, ITALY
Abstract: Digital farming as we see it has the potential to revolutionize agriculture, and bring
significant benefits for farmers and the society overall, as we need new ways to grow more food
more sustainably. In this study, Sustainable Farming and Digital Age, Certification as a Model for
a Sound Precision Agriculture and Impacts of Precision Agriculture and Certification Needs are
described.
Keywords: Digital farming, Sustainable Farming, Certification
Introduction
Smallholder agriculture still dominates the European rural economy, with 86% of EU farms
holding an area below 20ha (CEMA, November 2017).
Advanced agricultural machinery
solutions can help farm holdings, regardless of their size, to operate in a profitable, competitive
and sustainable manner. In particular, Precision Agriculture (PA) technologies holds great
potential for farmers in this regard. However, available economic evidence shows that there is a
strong link between the size of a farm holding and its income, with larger farms tending to have
higher income and investment capacity. Precision agriculture, a farming management concept
based upon observing, measuring and responding to inter and intra-field variability in crops, or to
some aspects of animal rearing, is a new frontier based on use of innovative technologies, such as
Global Navigation Satellite Systems (GNSS) and aiming at a rational adoption of decisions and
planned agricultural works well balanced between competitiveness and sustainability.
This concept has been simply explained as a way to apply the right treatment in the right
place at the right time”. As a consequence the world of agricultural machineries is quickly
changing with the evolution of modern, competitive and sustainable farming. If this is the scenario
for the mechanization of agriculture in the 3rd millennium, the principle to use innovative and
25
low-cost technologies to improve farming sustainability and competitiveness is a mandatory
choice for all farmers. That is why it is necessary to plan efforts and resources to train the majority
of farmers and more specifically all the youngsters, on the correct use of farming machinery, the
most advanced as well as the less modern but still at work machineries.
The challenge is therefore to combine Open Education Resources (OER) with user friendly
quality training materials available online. There is an increasing demand of online courses
targeted to work updated competences, as in the case of smart farming can be Internet of Things
(IoT) connected to on field good practices. These kinds of training needs are particularly requested
for use, management and controls of agricultural machinery and equipment, also not depending of
kind of agriculture, organic or conventional, due to a growing awareness to keep safe work place,
natural environment and food.
Correct and sustainable use of fertilizers and pesticides is one of the most important
farming issues, and training is fundamental to avoid risks and limit pollution as much as possible.
The most effective way to spread knowledge and competence on correct use of agricultural
machinery implemented with digital technologies is e-learning, since internet gives the opportunity
to provide high quality advice to a large number of users.
Sustainable Farming and Digital Age
The digital age in which we live needs responses consistent with global challenges and web
opportunities implementing effective quality teaching materials with existing online training and
education system such as Massive Open Online Courses (MOOC). Youngsters are very skilled
with internet technologies and they are the main target users, but it is also useful to combine well
developed e-learning materials, self-paced and also facilitated/instructor led, with work sharing,
work-shadowing and internships with the contemporary advice of experts and adult farmers. This
approach addresses therefore most of farmers who can be involved in e-learning and combined
training.
A tutorial is also needed to explain how to use an online course and how the interactive
learning platform can support course delivery and even communication among participants,
through forums and social networks. Nevertheless there is a lack of structured training platforms
and interactive teaching materials to get adequate competences from a basic use and maintenance
up to more detailed skills, based on the European Qualifications Framework (EQF), for farmers,
26
students, technicians and workers also to induce good pratices, information and training on safe
and sustainable use of digital agricultural machinery.
Concretely, it means applying new
technologies such as data science, advanced sensors in the field and flying drones, digital
communication channels, and automation on the field. This way more and more farmers have
access to better insights to take more optimal decisions, drive up yield, reduce using pesticide etc.
Current state of researches indicates that to reduce significantly diffuse pollution during
pesticides application, the emphasis must be placed on tools and methods for agricultural
professionals.
Three key aspects are involved:
• to optimize the agronomic decisions
• to control the precision of the applications
• to record the work performed.
All recent studies show a significant potential for reducing pesticide use: the emphasis
should be placed on tools and methods to optimize decisions concerning the use of pesticides and
the quality of applications. Guidance systems will be particularly explained as drivers for Precision
Agriculture linking farming with technological competences. They can be used by all kinds of
equipment (e.g. tractors, combine-harvesters, sprayers, planters…) and as part of a broad range of
different agricultural applications.
Guidance systems focus on precise positioning and movement of the machine with the
support of a Global Navigation Satellite System (GNSS).
Guidance Systems (GS) enable:
• Field digitalisation
• Automatic steering
• Precise machine movement between plant rows
• Precision drilling and sowing
• Precision spraying
• Mechanical weeding
27
Certification as a Model for a Sound Precision Agriculture
A certification system based on measurement of good practice performances can have many
benefits for all stakeholders offering an added value to manufacturers, a quality guarantee to
consumers and an effective tool for farmers to make the proper choice for their investment.
Besides, it offers the public sector a strong and effective tool to identify the best
technologies in order to focus better subsidy policies and extension efforts. With this approach we
can assess that the certification of digital farming practices as well as technologies can be effective
even for education and training at different levels. At present time the main problem is that public
extension services as well as many education courses do not offer an up-to-date program
concerning digital farming and related technologies. In order to provide farmers with effective
knowledge it is highly recommended to certify technologies and offer a ranking based on the
improvement they can offer. Furthermore the certification of farming practises and technologies
can be useful for a sustainability index. Actually the only well established certification system for
these technologies is the ISOBUS providing for a unique dialogue system between tractor and
implement.
The weak point of ISOBUS is the cost and the fact that it is common in expensive and large
machines and not available on all small scale farming machines, while on the other hand it is a
user friendly system. It is well known that digital farming makes every operation in agriculture
more effective and based on the real needs of crops. In this frame a reduction of the use of
chemicals, water for irrigation and other inputs is to be expected when compared to traditional
agriculture. Furthermore even the quality of agricultural products will be better because of the
reduction of chemical residuals etc. As an example of index we can compare the use of agricultural
machinery in a frame of precision farming rather that in a traditional system. The less chemical
being used the higher ranking in the indexing of the process. The same can be done for all other
inputs.
Certification is usually based on specific check lists in order to establish analytic
parameters that will assess the real performance. In other words a minimum requirement could be
a traditional crop protection practice that using a fixed amount of a certain fertilizer or pesticide
might provide the best results for the crop.
28
On this basis every technology that will allow the same result or better with a reduced use
of whatever inputs, such as fertilizers, pesticides, fuels, etc. will get a better evaluation and ranking
in the certification process. Certification will assess the good final result with a reduced input
based on less chemical and/or a less intense use of machines compared to present practices.
Certification will then set up an index with a minimum requirement and a progressive better value
according to the results.
Precision Agriculture technologies are able to identify clearly the real need of crops at very
local level and optimize the inputs only on these locations. In other words fertilizers and pesticides
will be used only where it is really needed and not on the whole crop area. Furthermore extension
activities will be focused on making farmers aware of the benefits of new technologies and on the
use of the certification as a tool to identify the best technology to be used in their farm.
This system will allow farmers to have a clear idea on the effectiveness related with the
use of a certain technology in terms of less work inputs, better environmental conditions and
expenses.
The aim will be focused on demonstrating that the digital technologies used can have clear
benefits:
Environment: less fertilizers and pesticides will be used,
Quality of production: less chemicals = less residuals, so more healthy food products,
Quality of life: humans will be less exposed to chemicals,
Market: economic benefits for farmers and consumers.
All these parameters can be a framework of reference for certification and value of the
certification to a clear index assessment. The same process can be followed to reduce/optimize the
parameters (i.e. the use of water for irrigation). In addition their products can have a clear
traceability process (i.e. protocols based on DLT) of inputs being used in order to get a higher
value on markets. All these issues supporting a more sustainable farming will be clearly identified
and measured in relation to a set of indicators certifying a sustainability rating.
29
Impacts of Precision Agriculture and Certification Needs
Impacts on different regional levels based on the foreseen changes can be evaluated and measured,
and the outputs described.
The overall desired impact is to foster enrepreneurial and professional handling of
innovative farming means by farmers, strengthening their business and role in the market and in
society with effect on more sustainable development in rural areas and effective reduction of water,
soil and air pollution and GHG emissions.
Conclusion
Entrepreneurial implementation of this kind of financing is a great challenge. Therefore, it is also
highly important to raise awareness for the importance of this topic amongst all rural stakeholders
and authorities. There is a strong need for further awareness and recognition at regional and
national level. By offering knowledge on existing practical successful experiences of innovative
farming based on Precision Agriculture at European level the project will deliver very useful inputs
for this important innovative trend for a more sustainable development in rural areas and launching
a broad discussion and dialogue as central basis for well led conceptualization and implementation
of innovative digital farming, starting from the State-of-the-Art of this very recent development
and ongoing experience for different crops and agricultural sectors.
The impact can be evaluated at different levels. At local level:
i) Exchange and development of Precision Agriculture on farms, with specific reference to
IoT and Apps for spraying machinery.
ii) Practical case studies inspring feasible introduction of Precision Agriculture on farms
iii) Rising awareness for innovatve trends introducing planning digital farming solutions
for agricultural machinery on farms, also improving quality of life in rural areas.
At national level:
i) Further development of the project results in collaboration activities with national
authorities and scientific institutions, keeping a strong focus on good practices emerging from the
case studies.
30
ii) Future inputs of practical ways of realizing concepts and training materals showed in
case studies in benchmarking from different countries and embedding transnational ideas and
experiences in the field of Precision Agriculture to different international and national contexts.
iii) Further national work in this field with analysis of most common problem solving
supporting evolving trends for precision agriculture.
At European Level:
i) Awareness on importance and feasibility of Precision Farming introductory planning as
important part of sustainable rural development and employment social inclusion as well as new
opportunity for economic development in rural areas
ii) Exchange among experts linking theory and practice at European level aiming at further
convergence of European nations and regions for planning measures of sustainable farming based
on Precision Agriculture.
iv) Development of further relationships at European level involving Turkey in the field of
Precision Agriculture for a more sustainable soil, water and air, fostering exchange of experiences.
REFERENCES
[1] https://www.cema-agri.org/publications/8/download
[2]https://www.slideshare.net/ChristianMStracke/ 20180713-mooq-conference-in-athens-mooq-andthe-quality-of-moocs-how-it-started-and-continues-stracke
[3] https://www.gsa.europa.eu/european-gnss/what-gnss
[4] https://www.aef-online.org/the-aef/isobus.html
[5]
https://www.worldgovernmentsummit.org/api/publications/document?id=95df8ac4-e97c-6578-
b2f8-ff0000a7ddb6
[6] Balsari P. et al. Developing Strategies to Reduce Spray Drift in Pneumatic Spraying in Vineyards:
Assessment of The Parameters Affecting Droplet Size in Pneumatic Spraying, Oct. 2017 Science of The
Total Environment
31
[7] http://www.agriprecisione.it/wp-content/uploads/2010/11/ general_introduction_to_ precision
agriculture.pdf
[8]
https://www.oliverwyman.com/our-expertise/insights/2018/feb/agriculture-4-0--the-future-of-
farming-technology.html
[9] European Commission, Internal Market, Industry, Entrepreneurship and SMEs Industry 4.0 in
Agriculture: Focus on IoT Aspects, July 2017
[10]https://www.euractiv.com/section/agriculture-food/infographic/farming-4-0-the-
future-of-
agriculture/
[11]
https://ec.europa.eu/eip/agriculture/sites/agri-eip/files/eip-agri_focus_group_on_precision_
farming_final_report_2015.pdf
[12] European Parliament, Precision Agriculture in Europe. Legal, Social and Ethical Considerations.
European Parliamentary Research Service Author: Mihalis Kritikos
32
Facebook Games Applications
Babaev Vladimir Yandashevich
Master of Arts in International Business and Management, Lecturer of Tashkent State University
of Uzbek Language and Literature Named After Alisher Nava'I, Taskent City, UZBEKİSTAN
Abstract: This article sheds light on how games applications increased its popularity using social
network platform such as Facebook. What is more, this piece of writing reflects rapid evolution of
particular games such as “Farmville”, “Pet society”, which using Facebook API. In addition, the
article provides information what technologies, policies, protection measures; Facebook takes to
protect users’ personal information, “OAuth 2.0 protocol”, in particular. Additionally, the article
provides information concerning benefits Facebook and its users get from using particular games
apps, challenges they are facing. Finally, the article gives some recommendations how Facebook
and its followers can cope with these challenges.
Keywords: Online Social Networks; Web 2.0; OAuth 2.0 Protocol; Application Programming
Interface
Introduction
Due to the rapid development of Web 2.0, popularity of Online Social Networks (OSNs)
dramatically increased for the last two years. According Scott Golder et al. in the USA alone,
number of undergraduate students using OSNs every day reached number of 90%. (Scott Golder
et al. 2007). No wonder that, OSNs such as Facebook is the most visited website in the internet.
(Comscore, 2008). Looking at the UK, 10% of all connections to the internet are to OSNs. What
is more, popularity of OSNs outweighed even pornography websites (R. Goad, 2009).
Facebook has created a unique digital environment for third party developers to design
various applications running on Facebook platform, in order to enhance number of users on its
page, which constituted more than 200 million users in western countries alone, while China users
hit the number of more than 300 million. (Cosenza, V, 2009). Games applications designers found
opportunity to use Facebook as a gaming platform quite lucrative.
33
Facebook is the Best Platform For Games Apps
Despite the fact that number of SNSs largely contributed to the expansion of games Facebook is
the one of the best platforms, where games applications has the biggest impact. For instance,
“Farmville”, which is simulation farming management created by Zinga Company, designed for
Facebook platform, currently has 83,131,550 active users. Whereas, “Mafia Wars” the game in
role-play genre created by the latter developer’s number of players hit 25,225,819. 5. (Ines Di
Loreto, 2010).
Many games designed for Facebook based on browser, and, as a result do not require
further installation (Ines Di Loreto, 2010). This fact makes it easy for users to navigate through
the game, not spending a lot of time on installation process.
In December 2009, 208 apps were the most downloaded games applications in the
Facebook official website. One of the most popular casual games was “Solitaire” apps, with
600,000 active users. Creation of “casual games” is not incidental, these kinds of genres designed
specifically for users who position themselves as “casual gamers”, rather than “gamers”. Taking
into consideration, the casual gamers and specific niche that these users comply together,
Facebook’s engineers designed quite new game category “Just for Fun”, which put together
players regarding themselves “casual gamers”. (Ines Di Loreto, 2010).
These actions taken by Facebook, shows its great adaptability to users’ preferences and
tastes in games apps. Casual games in its nature do not necessarily have complicated gameplay
and designed mainly for entertaining and relaxation. (Ines Di Loreto, 2010). As Ahn, L. pointed
out, user spend much of their time on games, which successfully designed. (As Ahn, L, 2006).
Bearing his words in mind, number of games designers trying their best to create something great,
thus luring more users to install their apps.
Facebook API As Incitement For Third Party Designers
Owing to Facebook Application Programming Interface (API), third party games designers can
easily integrate their apps with Facebook platform. Thus, “Texas Holdem Poker” apps are one of
the most successful games apps on Facebook with more than 1,7 million users on daily basis
(Adonomics, “Top applications.”).
34
Since 2007, thousands of apps were brought on Facebook, using its API. Various apps themes
were introduced, such as presenting gifts, casual talking among friends, bodily movements,
suggestions, etc. (M. Gjoka, et al. 2008).
Sharabi (A. Sharabi, 2007) identifies apps according to their social intentions:
“Self-Presentation Tools”, which is how people distinguish their authenticity with
regard to their best movie or present state of feeling.
“Collective Identity Formation” is asking people how they define other peoples’
authenticity - Pleasant or Unpleasant games apps asking users to choose any adjective to depict
their friend.
“Phatic Communication Tools” is sending “gifts”, “hugs”, etc. in order to sustain
“social contact” among users.
Looking in more details at “Hugs” application (http://apps.facebook.com/huggees), users pick up
any kinds of “hugs” which they intend to send “Fuzzy Hug” or “Friendly Hug” , whereas,
addressee receive a “hug” together with a picture panda, dog any funny animals. (A. Nazir, 2008).
It is clear that games apps designers, together with Facebook engineers, aim to strengthen
relationship among “friends”, by using quite unobtrusive communication tool.
The “Oauth 2.0 Protocol” As Additional Safety Measures
Facebook quite masterly uses information management to protect their users, which is their main
priority. Thus, Facebook uses “OAuth 2.0 protocol”, in order to reduce information transferring
between users and third parties’ apps designers, who can intentionally violate users’ private
information, can distribute it to other parties, without user consent. This protocol provides
additional protection authentication level, when user trying to install apps on his or her Facebook
profile, application that is going to be installed asks users’ permission to his or her information
access, which is kept on Facebook. (N Wang et al, 2011). Below there is Figure 1 demonstrating
how “OAuth 2.0 protocol”, operates.
35
Figure 1. The OAuth 2.0 Protocol 4
Source: N Wang et al, 2011.
This information usually concerns user’s name, gender, friends’ list, avatar, etc. Such kind
of information refers to “basic information”, and if application trying to acquire data, which is
beyond the “basic information”, this application needs to prolong user’s consent. Generally,
additional consent refers to extended information of user, which include his or her personal
contacts, friends’ personal information, photos, videos, etc. (N Wang et al, 2011). Figure 2 depicts
claim to this consent.
36
Figure 2. Request for Extended Permission
Source: N Wang et al, 2011.
Information control is user’s prerogative Users in turn, can protect their personal
information by adapting their “privacy settings”, in order to restrict or limit access to their private
data to other users in the cyberspace. By changing private settings, user can control general
information his or her friend can use to locate him or her on Facebook. Secondly, regulate who
will able to inspect their sharing content. Last, but not least, users are able to modify list of people
who is blocked. (N Wang et al, 2011). Figure 3 demonstrates the interface of “privacy settings”.
37
Figure 3. “Privacy Settings”
Source: N Wang et al, 2011.
Bearing in mind the information above, it is clear, that Facebook interface and digital
environment are quite flexible. Users can modify, number of settings to their needs and tastes.
Undoubtedly, users control and manage their information according to their preference; they
choose whom to give access to their personal content, whom to make friends with, whom to
blocked. Generally, Facebook grants “carte blanche” to user.
“Playfish’s” “Pet Society” As A Great Socializing Tool
Facebook games applications represent not entertainment and relaxation solely, these apps bring
people together, consolidate the unity inside the friends’ group. Games apps unify not just user’s
friends, whom he or she can meet on a regular basis but distant relatives, who live in a faraway
country, and there is rare possibility to club together more often.
This great opportunity represents “Playfish”, company headquartering in UK, founded in
2007 by, Sami Lababidi, Kristian Segerstrale, Shukri Shammas and Sebastien de Halleux, and,
and now is in Electronic Arts’ possession (www.playfish.com/?page=company).
38
“Playfish” for quite a short period made a spectacular revenue and attracted millions of
users. “Playfish” designs its games mainly for Facebook platform attracting above 11 million users
every month. Currently, 5 out of 7 “Playfish” company’s games are leading the top 25 games
applications based on Facebook platform. “Pet society”, which is the biggest success of the
“Playfish” company, and the first application that reached 1 million players on Facebook alone.
(L. Rossi, 2010).
“Pet society” success phenomenon is quite simple to distinguish. Firstly, this game
designed for all ages, genders; no matter do they have their own pets or going to acquire any.
Players find this game quite interesting, non-obtrusive, and fun. Figure 4 illustrates screenshot of
“Pet society” apps.
Figure 4. “Pet society”
Source: www.free2play.com, 2013
The gameplay is quite simple: players looking after their pets, feeding, playing with them,
etc. “Pet society” has two modes “Single” and “Multiplayer”. Players within the friends’ network
can visit each other’s pets, doing various activities such as cleaning, giving food, etc. Players,
getting coins for each visit of their friends’ pets visit. In order to obtain more coins, players should
extend their friends’ network, more friends, higher revenue, and as a result greater achievement.
39
With more coins in their pocket, pets’ owners are able to buy more stuff for their pets
(www.facebook.com/petsociety).
Taking into consideration information mentioned above, it is obvious that players to obtain
desired coins, actively reviewing their current friends list, and trying to become friends with as
many people as possible. These potential friends, who can largely contribute to player game status,
come not just from close circuit, but even people they are working with and ex-schoolmates, whom
players did not consider as friends. (L. Rossi, 2010)
Due to games requirements in obtaining coins, players should extend their friends list, to
be able to get more coins. Usually players, create separate group of “friends” which are people
users interacting during the game process, rather than hanging out together. Eventually, “friends
for social gaming” can transform into real life friends.
Owing to unique Facebook digital environment, every action, committed by player, all his
achievement during the game process, people they become friends with, instantly display on
player’s profile. This fact is a challenge for other users, who is involved in the same game, play
more often, becoming friends with larger number of people. What is more, player should maintain
this created relationship; otherwise, they can lose their superiority in the game. (L. Rossi, 2010)
Facebook demonstrates effective information control and marketing strategies. Firstly, they
are cooperating with third party developers, extending their influence area among games designers
and technical community. Second thing is that owing to Facebook people are always in touch with
each other and it does not matter what brings them together, game or any other occasions. The
most important thing is that people gathering, sharing their success, achievements, etc., no matter
in a real-life world or in the cyberspace.
Foul Play of Games Apps Designers
However, despite Facebook assurance of users’ personal information protection, there some flaws
in games apps designed for Facebook platform. According to a Wall Street Journal’s finding,
number of popular games apps in Facebook, pass users’ personal information including age,
photos, friends list to third parties, such advertising agencies, data collection centers, thus violating
users’ privacy, without their awareness of this actions taking place. Among privacy offenders is
“Zynga Company”, “Farmville” game apps creator, which accused of transferring users’ friends’
40
details to third parties organizations. Currently, “Farmville”, is not accessible for Facebook users.
“LOLapps” and “Family tree” apps have been passing users’ ID to “RapLeaf” data center, a Wall
Street Journal have found. All companies mentioned denied the fact of privacy violation, and
declared all actions were unintentional, and their main priority was and still users’ privacy
protection. After findings publication, Facebook shut down number of apps depicted in
investigation. As Facebook authorities claimed, they are taking all necessary steps to prevent users’
personal information leakage. Nevertheless, it is still unclear what steps, they are taking and how
they can stop this information leakage. It is true, that due to Facebook privacy regulations, games
designers restricted players’ personal information transmission to other parties, even though, with
user’s approval. However, it should be noted, that Facebook highly relied on varies apps created
by third parties’ designers, because apps are great tool in extending Facebook network. Currently,
70 percent of Facebook users utilizing various apps on monthly basis (E Steel, G Fowler, 2010).
Despite, shutting down apps, which is not reliable, trustworthy and do not complying with
privacy regulations, Facebook cannot stop them operating all at once. What is more, it is not clear
how many other apps and its designers violate users’ privacy.
Conclusion
Undoubtedly, Facebook platform one of the most popular social network in the world, with
millions of users all around the world. Facebook is not just the platform where are people are
getting together, it is more than that, it is a quite new “digital world”, where every current or
potential user can find something extraordinary for themselves, and make something they really
want into reality, even impossible from the first sight. It is out of the question that among the vast
majority of apps available in Facebook, games apps has its own niche. Games apps for quite a
short period become extremely popular among Facebook users. It is evident, that games apps
encroach users’ privacy, and there a long way ahead for Facebook to protect personal content
millions of its users.
On the other hand, Facebook developing new strategies, policies, to prevent users’ private
information dissemination to other parties. Moreover, Facebook grants more options to users in
setting up their privacy settings within their profiles. Additionally, Facebook lets people manage
their own information, according to their personal preferences. The best strategy for Facebook to
prevent its users’ privacy violation is to limit number of games creators using its platform and
41
decrease collaboration with third parties’ apps designers. Finally, to start designing more their own
apps themselves, thus ensuring users’ privacy protection and less reliability on other apps creators,
who intentionally or unintentionally can undermine users’ confidence.
It is still unclear, should people share their personal details with any other social networks,
in order to maximize protection of their accounts. Nevertheless, it is out of the question, that any
social networks and apps cannot substitute face-to-face socializing.
REFERENCE
[1] A. Nazir, S. Raza, and C.-N. Chuah, “Unveiling Facebook: A Measurement Study of Social
Network-Based Applications,” in IMC ’08: Proceedings Of The 8th ACM SIGCOMM Conference on
Internet Measurement. New York, NY, USA: ACM, 2008, Pp. 43–56.
[2] A. Sharabi, “Facebook applications trends report,” November 19th, 2007. [Online]. Available:
http://no-mans-blog.com/2007/11/19/facebookapplications-trends-report-1/
[3] Ahn, L. v. Games with a Purpose. Computer, 39, 6 (2006), 92-94.
[4] "About Playfish Company & Founders" http://www.playfish.com/?page=company
[5] Adonomics, “Top Applications.” [Online]. Available: http://adonomics.com/leaderboard.
[6] Cosenza, V. World Map of Social Networks. Vincos Blog, 2009. http://www.vincos.it/worldmap-of-socialnetworks.
[7] Comscore. http://www.comscore.com/, Apr 2008.
[8] E Steel, G Fowler- The Wall Street Journal, 2010 Facebook in Privacy Breach - terriau.org,
http://terriau.org/blog/postings/20101018%20Facebook%20in%20Online%20Privacy%20Breach%3B%2
0Applications%20Transmitting%20Identifying%20Information%20-%20WSJ.pdf
[9] Ines Di Loreto, Abdelkader Gouaich. Social Casual Games Success is not so Casual. RR-10017.
[10]
2010,
pp.001-011.
<lirmm-00486934>
http://hal-lirmm.ccsd.cnrs.fr/lirmm-00486934/
document.
[11] L Rossi Playing Your Network: Gaming in Social Network Sites. Available at SSRN 1722185,
2010 - papers.ssrn.com http://www.digra.org/wp-content/uploads/digital-library /09287.20599.pdf
42
[12] M. Gjoka, M. Sirivianos, A. Markopoulou, And X. Yang, “Poking Facebook: Characterization
of Osn Applications,” in WOSP ’08: Proceedings Of The First Workshop On Online Social Networks. New
York, NY, USA: ACM, 2008, Pp. 31–36.
[13] N Wang, H Xu, J Grossklags - Third-Party Apps On Facebook: Privacy And the Illusion of
Control
Proceedings
Of
The
5th
ACM
Symposium,
2011-dl.acm.org,
http://personal.
psu.edu/nzw109/papers/Wang_chimit_2011.pdf.
[14] R. Goad, “Social Networks Overtake Adult Websites,” Hitwise Intelligence Online, 2009.
[15] Scott Golder, Dennis Wilkinson, and Bernardo A. Huberman. Rhythms of Social Interaction:
Messaging within a Massive Online Network. In International Conference on Communities and
Technologies, 2007.
[16] www.free2play.com
[17] https://www.facebook.com/petsociety
43
A Note on the Examination of Portugal’s Hotels Performance - a Proposal for a New
Perspective’s Approach
Nuno Mendes FREITAS, José António FILIPE
Information Sciences, Technologies and Architecture Research Center
Instituto Universitário de Lisboa, PORTUGAL
Abstract: The main objective of the present research paper is to give an exploratory vision on a
new form of examining hotels’ performance of the Portuguese hotels’ market. This approach is
based on the experience of a consultant in this area and is made in terms of revenue and occupancy,
crossing then operational data with financial results. This analysis focuses on a recent period, since
2010 to 2017, involving the years since before and until after the big boom of tourism in Portugal.
This paper exposes available data about hotels operational results and compares these results with
hotels financial results. In terms of the evaluation of a hotel’s performance, the revenue per
available room (RevPAR) is regarded by the tourism industry and INE (the Portuguese National
Statistics Institute) as the most important measure, once it compares hotels among themselves and
also compares regions. An important aspect of this study is highlighting the limitations of
RevPAR.
Keywords: Revpar, Financial Results, Hotels Performance
Introduction
The present study is based on the operational and financial data available for the period between
2010 and 2017 regarding to Portuguese hospitality. For several decades, the lodging industry has
used RevPAR (revenue per available room) as a key indicator to evaluate a firm’s performance
and to make investment decisions [1]. RevPAR is one of the most recognized and used
performance measures in the hospitality industry, providing general market trends and some
revenue indications. There are some pitfalls to be aware of when analyzing a hotel’s performance
based solely on RevPAR [2]. RevPAR only measures revenues; it ignores expenses and does not
44
capture or report profitability. Thus, RevPAR is a rather limited measure for measuring the hotel’s
performance [3]. This indicator is calculated by dividing a hotel’s net room revenue (considered
after discount and sales taxes and net of breakfast and other meals) by the total number of available
rooms and by multiplying a hotel’s average daily room rate (ADR) by its occupancy [2].
Hotels, with significant revenue-generating departments rather than the rooms division, by
focusing only on room revenue do not provide a complete or accurate picture of the entire hotel
property’s performance [3].
For the analysis of the hotel’s performance, other complementary metrics should be used to
measure the financial performance, by crossing the financial result with the number of occupied
rooms or available rooms. This will allow hotel decision makers to take into account the costs of
the operation in the creation of the room price, and to realize how their operating decisions and
cost structure reflect the net revenue. RevPAR is an important indicator when comparing the hotel
performance with the one of competitors and with the market. Anyway, it is not the best indicator
to analyze the operational performance of hotel units. A greater value in the RevP AR to the market
or competition is not necessarily a sign that the hotel is performing better than the others, as each
hotel company has different operating structures with different costs.
The calculation of RevPAR
Rooms’ revenue per available room (RevPAR) measures the rooms’ revenue yield that a property
achieves in terms of the rooms available in the property for a given period. It includes the influence
of occupancy and ADR.
RevPAR is calculated as follows:
RevPAR = Total Rooms Revenue / Number of Rooms Available
The number of rooms available or the number of the rooms’ nights available is based on the
calculations of weighted average number of bedrooms in the hotel, multiplied by the total number
of nights in the period under the analysis.
Total rooms revenue is the number of rooms’ nights sold in the period of analysis.
45
The Portuguese Tourism and Hotels Market
In this market, there was recently a recession in Portugal, involving the period between 2010 and
2014 most considerably, but having started with the global financial crisis in 2007-2008.
TABLE I.
GDP GROWTH RATE – PORTUGAL. UNIT: EURO – MILLIONS
GDP Growth Rate
2010
1.90
2011
1.83
2012
- 4.03
2013
- 1.13
2014
0.89
2015
1.82
2016
1.93
2017
2.80
Source: BdP.
In recent years, Portugal had a significant growth in terms of tourism. In 2017, 12.7 million
tourists visited Portugal, representing 3.4 billion euros in terms of hotels’ revenues. Tourism is one
Portuguese key sector, having a great importance for the Portuguese economy with a relative
weight of 13.17 % in Portuguese GDP in 2017. The number of accommodations in Portugal,
according to the National Statistics Institute (INE), is 5840 in 2017.
TABLE II.
Travel and Tourism Account as a Percentage of GDP. Unit: Euro - Millions
Year
2010
2011
2012
2013
2014
2015
2016
2017
% of GDP
6.31
7.16
7.78
8.78
10.10
10.75
11.57
13.17
Source: INE.
46
Literature Review
Many observers, particularly those in the financial community, view the hospitality industry
generally as being a risky business for owners and investors. This vision can be explained by
considering much of the investments risk stems in the fundamentals of this business, namely,
room-rate and occupancy fluctuations [4]. Several research papers in the literature show the
importance of RevPAR as a financial performance indicator ([5]. Higgins interviewed industry
analysts and managers, regarding the importance of RevPAR [6]. These interviews indicated
RevPAR as the most widely used measure, internally and externally, for hotels performance
analysis and accepted by lodging firms as a benchmark. A limited amount of research discussed
and investigated the relationship between the lodging firm performance and RevPAR [7]. . Brown
and Dev highlighted two key limitations of RevPAR: it does not include revenue from food and
beverage and other departments; and it does not take into account costs that are incurred to provide
the requisite service level (e.g., special amenities such as a spa or additional guest-service
employees such as a concierge) [8].
Gross Operating Profit per Available Room (GOPPAR) has become popular as an
important alternative performance measure because it resolves limitations of RevPAR [9]. It
provides a deep indication of a hotel’s profitability by taking into consideration management
control and efficiency, and eliminating, to a certain extent, the potential advantage of a small hotel
on this analysis.
In addition, GOPPAR offers an overall more robust performance measure, especially when
comparing the financial performances of hotels with different sizes or in different markets [2].
One of the issues in lodging finance is the ratio of debt to equity [10]. According to [11],
the hospitality industry is traditionally confronted with a high need for financial capital to invest
in fixed assets such as land, building, and equipment, and since debt is relatively cheaper than
equity, it has been widely used as a source of capital to fund investments.
Methodology
a. The Study
The analysis of data in this study has two stages. First, the study is conducted considering the
hospitality statistical data issued by the Portuguese National Statistics Institute. The number of
47
occupied rooms and available rooms is calculated and then RevPAR and ADR from 2010 to 2017
are also calculated.
Second, financial reports of available data, assessed from the Central balance sheet database of the
Portuguese Central Bank (Banco de Portugal), are used as the relevant financial information for
the examination of the perspective of financial performance per available room and per occupied
room.
b. The Data Collection
The study collected data from two main sources.
Financial data are collected from the Portuguese CAE 551 – Hotels and similar accommodations.
The operational data from INE statistics used in this study are the tourism statistics, having been
this information collected by INE, from a survey in hotels and other similar establishments.
c. Corporate performance measures
Return on assets (ROA) is the net income divided by total assets. It reflects the ability of a firms’
management to generate profits from firms’ assets. Hotel managers have to ensure great profits in
order to cover the high assets costs and the related fixed costs. These profits have to be coherent
with hotel investments. The ROA is retained to evaluate hotel profits enclosing all provided
services and activities.
ROA = (Net Income/ Total assets) X 100
Return on equity (ROE) is used to measure profitability of companies, and is the net income
divided by total equity, measuring the firm efficiency to generate profits from shareholders
investment. The company’s objective is the maximization of shareholder’s value to have an
optimum mix of debt and equity without compromising ROE.
ROE = (Net Income / Total equity) X 100
Data Analysis
RevPAR variability is measured by the mean of the absolute values of the annual changes in
RevPAR. Table III presents the mean of the annual measures during the period from 2010 to 2017.
It is noted that the RevPAR variability is growing during the period between 2012 to 2017.
48
TABLE III.
REVPAR. UNIT: EURO
Year
Revpar
2010
2011
2012
2013
2014
2015
2016
2017
28,27
29,39
28,47
30,20
33,01
37,60
43,23
50,26
Variabili
ty
4%
-3%
6%
9%
14%
15%
16%
Source: BdP.
When analyzing the ROA, we verify that it is inconsistent over the years under analysis,
comparatively to RevPAR that had a slight decrease from 2011 to 2012 and returned to a constant
growth from 2012 to 2017.
Assets grew during the years under analysis, and the negative variation was verified in net
income.
TABLE IV.
ROA
Year
ROA
2010
2011
2012
2013
2014
2015
2016
2017
2.35%
0,89%
0,50%
1,87%
3,27%
5,15%
5,38%
7,34%
Variabili
ty
-62%
-44%
274%
75%
57%
4%
36%
Source: BdP
From 2010 to 2104, ROE has always had a negative value, due to the Net results of hotel
companies. Comparing to RevPAR, the variability is not constant. RevPAR shows a growth
between 2013 and 2017; ROE only grows in 2016 and 2017.
49
ROE.
TABLE V.
Year
ROE
2010
2011
2012
2013
2014
2015
2016
2017
-5,26%
-10,96%
-15,90%
-10,82%
-5,21%
1,56%
2,07%
7,61%
Variabili
ty
-108%
-45%
32%
52%
130%
33%
268%
Source: BdP.
When analyzing EBITDA or Net Income per occupied room, we observe that Revpar grew from
2012 to 2017 and had a constant positive variation. However, we do not observe the same trend
when analyzing the values presented in Table VI and table VII.
TABLE VI.
EBITDA PER OCCUPIED ROOM. UNIT: EURO
EBITD
A
12,59
4,90
2,84
10,45
16,65
24,57
25,39
33,84
Year
2010
2011
2012
2013
2014
2015
2016
2017
Variabi
lity
-61%
-42%
268%
59%
48%
3%
33%
Sources: BdP.
TABLE VII.
NET INCOME PER OCCUPIED ROOM. UNIT: EURO
Net
Income
-9,76
-17,61
-22,95
-14,71
-6,80
2,11
Year
2010
2011
2012
2013
2014
2015
50
Variabi
lity
-80%
-30%
36%
54%
131%
Net
Income
-9,76
-17,61
3,07€
11,56
Year
2010
2011
2016
2017
Variabi
lity
-80%
45%
277%
Sources: BdP.
Discussion, Conclusion and Limitations
We can verify that complementary ratios should be used to achieve and control financial results.
The used ratios do not measure financial costs, only the operational costs of the hotel.
Another strand for the analysis is related to the capital structure used in hotels. Investments are
made using financing, being the financial autonomy always below 30% and for that reason, in
some years, the net income was negative. The use of the net income per available room can be a
starting point to measure the financial performance.
RevPAR is the dominant, and currently, the most used and accepted measure by tourism and
hotels, and, considering that, it is possible to understand how important is to search for alternative
performance measures.
REFERENCES
[1] J. Chen, Y. Koh, And S. Lee, “Does The Market Care About Revpar? A Case Study Of Five
Large U.S. Lodging Chains,” Journal Of Hospitality And Tourism Research, Vol. 35, No. 2, 2011.
[2] E. Younes And R. Kett, “GOPPAR, A Derivative Of Revpar,” HVS International, No. March,
2003.
[3] S. Lee, B. Pan, And S. Park, “Revpar Vs. GOPPAR: Property- And Firm-Level Analysis,”
Annals Of Tourism Research, Vol. 76, No. September 2018, 2019.
[4] J. A. Ismail, M. C. Dalbor, And J. E. Mills, “Using Revpar To Analyze Variability,” Vol. 24,
No. 4, 2002.
[5] M. Gallagher And A. Mansour, “Analysis of Hotel Real Estate Market Dinamics.” American
Real Estate Society, 2000.
51
[6] S. M. Higgins, “Revpar Still King, But GOPPAR on The Rise,” Hotel & Motel Management,
Vol. 221, No. 1, 2006.
[7] A. R. Elgonemy, “The Pricing Of Lodging Stocks: Reality Check.” Cornell Hotel And
Restaurant Administration Quarterly, 2000.
[8] J. R. Brown And C. S. Dev, “Looking Beyond Revpar: Productivity Consequences of Hotel
Strategies,” Cornell Hospitality Quarterly, Vol. 40, No. 2, 1999.
[9] S. Lee, B. Pan, And S. Park, “Revpar Vs. GOPPAR: Property- And Firm-Level Analysis,”
Annals Of Tourism Research, Vol. 76, No. September 2018, 2019.
[10] A. R. Elgonemy, “Debt-Financing Alternatives,” No. June, 2002.
[11] Y. Li And M. Singal, “Capital Structure İn The Hospitality İndustry: The Role of The AssetLight And Fee-Oriented Strategy,” Tourism Management, Vol. 70, No. August 2018, 2019.
[12] BP STATS – Banco De Portugal 2010 To 2017, Lisboa.
[13] INE Instituto Nacional De Estatística 2010 To 2017, Lisboa.
52
A Comparative Case Study on Time Series Prediction
Anıl ÖZDEMİR, Furkan COŞKUN, Selim BALCISOY
Faculty of Engineering and Natural Sciences, Sabancı University, TURKEY
Abstract: A time series is a sequence collected at consecutive equally spaced points in time. The
basic idea behind the time series forecasting is the use of a model to estimate future values based
on previously observed ones. Traditionally, statistical methods are used to forecasting time series
however, Machine Learning (ML) algorithms have been also proposed as alternatives to statistical
methods in past decades. In this paper, we evaluate forecasting performance of different ML
algorithms and statistical methods on Turkey automobile sales. Recently, various of work has
claimed that traditional statistical methods dominate the ML solutions in terms of time series
forecasting. This study discusses different aspects of ML and statistical methods and compare their
performance on different time series.
Keywords: Time Series Forecasting, Machine Learning Regression, Statistical Models.
REFERENCES
[1] Brillinger, David R. Time Series: Data Analysis And Theory. Vol. 36. Siam, 1981.
[2] Brown, Robert Goodell. Statistical Forecasting For Inventory Control. McGraw/Hill, 1959.
[3] Whittle, P. (1951). Hypothesis Testing in Time Series Analysis. Almquist and Wicksell.
Whittle, P. (1963). Prediction and Regulation. English Universities Press. ISBN 0-8166-1147-5.
[4] Hannan, Edward James (1970). Multiple Time Series. Wiley Series in Probability And
Mathematical Statistics. New York: John Wiley And Sons.
[5] Box G, Jenkins G. Time Series Analysis: Forecasting and Control. San Francisco: Holden-Day;
1970.
[6] Brockwell, Peter J., Richard A. Davis, and Matthew V. Calder. Introduction to Time Series and
Forecasting. Vol. 2. New York: springer, 2002.
53
[7] Bauwens L, Laurent S, Rombouts JVK. Multivariate GARCH Models: A Survey. Journal of
Applied Econometrics. 2006;21(1):79–109.
[8] Makridakis, Spyros, Evangelos Spiliotis, and Vassilios Assimakopoulos. "Statistical And
Machine Learning Forecasting Methods: Concerns And Ways Forward." PloS one 13, no. 3 (2018):
e0194889.
[9] Gilchrist, Warren. Statistical Forecasting. Vol. 322. London: Wiley, 1976. [10] Zhang G, Eddy
Patuwo B, Hu Y M. Forecasting With Artificial Neural Networks: The State of The Art. International
Journal Of Forecasting. 1998;14(1):35–62.
[11] Kim, Kyoung-Jae. "Financial Time Series Forecasting Using Support Vector Machines."
Neurocomputing 55, no. 1-2 (2003): 307-319.
[12] Bontempi, Gianluca, Souhaib Ben Taieb, And Yann-Aël Le Borgne. "Machine Learning
Strategies For Time Series Forecasting." In European BusinessIntelligence Summer School, pp. 62-77.
Springer, Berlin, Heidelberg,2012.
[13] Adya M, Collopy F. How Effective Are Neural Networks At Forecasting And Prediction? A
Review And Evaluation. Journal Of Forecasting. 1998;17(56):481–495.
[14] Makridakis, Spyros, and Michele Hibon. "The M3-Competition: Results, Conclusions And
İmplications." International Journal of Forecasting 16, no. 4 (2000): 451-476.
[15] Hibon, Michèle, and Herman Stekler. The M-3 Competition: Statistical Tests of the Results.
INSEAD, 2003.
[16] Crone SF, Hibon M, Nikolopoulos K. Advances in Forecasting With Neural Networks?
Empirical Evidence From The NN3 Competition On Time Series Prediction. International Journal Of
Forecasting. 2011;27(3):635–660.
[17] Thiel, Christian, Johannes Schmidt, Arnold Van Zyl, And Eschmid. "Cost And Well-ToWheel İmplications Of The Vehicle Fleet CO2 Emission Regulation İn The European Union."
Transportation Research Part A: Policy And Practice 63. (2014): 25-42
[18] Hansen, James V., James B. McDonald, and Ray D. Nelson. "Time Series Prediction with
Genetic‐Algorithm Designed Neural Networks: An Empirical Comparison with Modern Statistical
Models." Computational Intelligence 15, no. 3 (1999): 171-184.
[19] Goodwin P, Lawton R. On The Asymmetry of The Symmetric MAPE. International Journal
of Forecasting. 1999; 15(4):405–408.
54
[20] Armstrong, J. Scott; Collopy, Fred (1992). "Error Measures for Generalizing About
Forecasting Methods: Empirical Comparisons" (PDF). International Journal of Forecasting. 8 (1): 69–80.
[21] Tipping, Michael E. "Sparse Bayesian Learning and The Relevance Vector Machine." Journal
of Machine Learning Research 1, No. Jun (2001): 211-244.
[22] Scikit-learn: Machine Learning in Python, Pedregosa et al., JMLR 12, pp. 2825-2830, 2011.
[23] Rokach, Lior; Maimon, O. (2008). Data Mining with Decision Trees: Theory and Applications.
World Scientific Pub Co Inc. ISBN 978-9812771711.
[24] Altman, N. S. (1992). "An İntroduction to Kernel And Nearest-Neighbor Nonparametric
Regression". The American Statistician. 46 (3): 175–185.
[25] Scho ¨lkopf B, Smola AJ. Learning with kernel: Support Vector Machines, Regularization,
Optimization and Beyond. The MIT Press; 2001.
[26] Breiman, Leo. "Random Forests." Machine Learning 45, No. 1 (2001): 5-32. [27] Yan, Xin,
And Xiaogang Su. Linear Regression Analysis: Theory And Computing. World Scientific, 2009.
[28] Weisstein, Eric W. "Binet-Cauchy Identity." CRC Concise Encyclopedia Of Mathematics (2nd
ed.), CRC Press, ISBN 1584883472 (2003): 228.
[29] Seabold, Skipper, And Josef Perktold. “Statsmodels: Econometric and Statistical Modeling
With Python.” Proceedings Of The 9th Python İn Science Conference. 2010.
[30] Hyndman, Rob J., And George Athanasopoulos. "8.9 Seasonal ARIMA Models." Forecasting:
Principles And Practice. Otexts. Retrieved 19 (2015).
[31] Pal, Sankar K., and Sushmita Mitra. "Multilayer Perceptron, Fuzzy Sets, Classifiaction."
(1992).
[32] Kedem, Benjamin, And Konstantinos Fokianos. Regression Models For Time Series Analysis.
Vol. 488. John Wiley & Sons, 2005.
55
Optimal PID-like Fuzzy Logic Controller Design for Ball and Beam System
O. Tolga ALTINÖZ, A. Egemen YILMAZ
Department of Electrical and Electronics Engineering, Ankara University, TURKEY
Abstract: Ball and beam system (BBS) is a benchmark hardware for designing control action. The
structure of the system is based on changing the angle of the beam so that the position of the ball
is changed. It is desired to move the ball to a reference position. In this paper Fuzzy Logic
Controller (FLC) is applied for this problem. Instead of conventional FLC, the derivative and
integral terms are integrated to the FLC, which is called as PID-like FLC. This controller has a
constant Fuzzy structure with variable parameters. The performance of the controller is based on
these parameters. Therefore, in this study, the parameters of PID-like FLC are optimized by using
three optimization algorithms; Genetic Algorithm, Particle Swarm Optimization, and Differential
Evolution. The performance of the controller is demonstrated on both simulation and hardware
environment. The performance of the optimization algorithm with respect to the obtained
performances are compared in this paper.
Keywords: Ball and Beam System, Particle Swarm Optimization, Genetic Algorithm, Differential
Evolution, Fuzzy Logic Controller, PID-like.
56
Adopting Machine Learning Algorithms for Cloud-Based Application Categorization
Çağatay ÇATAL1, Besme ELNACCAR2, Özge ÇOLAKOĞLU2, Bedir TEKİNERDOĞAN1
1
Wageningen University, NETHERLANDS; 2 Istanbul Kültür University, TURKEY
Abstract: Manual categorization of applications in software repositories such as SourceForge is
often time-consuming and error-prone. Automation of this process not only simplifies the daily
task of administrators but also helps project owners to add their projects into the corresponding
subcategory of the repository without any delay. In this study, we propose a cloudbased application
categorization system that applies machine learning algorithms to support the classification of
applications. The categorization system has a web-based client application to parse, process, and
submit the project source code, a web service which automatically performs classification of
applications into domain categories, and a cloud-computing platform which hosts the
categorization service. Several multi-class classification algorithms have been adopted including,
Artificial Neural Networks, Logistic Regression, Decision Jungle, and Decision Forest algorithms
to validate the effectiveness of the system in multiple case studies. The case studies were
performed on three public datasets generated based on 3286 Java applications of SourceForge
repository. Our study shows that the highest accuracy was achieved with Artificial Neural
Networks (ANN) algorithm. The resulting prediction model has been transformed into a web
service and then, deployed on the Azure cloud platform.
Keywords: Cloud Computing, Software Maintenance, Machine Learning, Application
Categorization, End-To-End Cloud System
REFERENCES
[1] E. Kalliamvakou, G. Gousios, K. Blincoe, L. Singer, D. M. German, and D. Damian, “The
Promises And Perils Of Mining Github,” in Proceedings Of The 11th Working Conference On Mining
Software Repositories. ACM, 2014, pp. 92–101.
[2] J. Howison And K. Crowston, “The Perils And Pitfalls Of Mining Sourceforge,” in Proceedings
of the International Workshop on Mining Software Repositories (MSR 2004). IET, 2004, pp. 7–11.
57
[3]SourceForge. (2018) Saourceforge. [Online]. Available: https: //sourceforge.net /directory/
os:windows/
[4] C. Catal, S. Tugul, and B. Akpınar, “Automatic Software Categorization Using Ensemble
Methods And Bytecode Analysis,” International Journal of Software Engineering and Knowledge
Engineering, vol. 27, no. 07, pp. 1129–1144, 2017.
[5] K. Tian, M. Revelle, And D. Poshyvanyk, “Using Latent Dirichlet Allocation For Automatic
Categorization Of Software,” in Mining Software Repositories, 2009. MSR’09. 6th IEEE International
Working Conference On. IEEE, 2009, Pp. 163–166.
[6] M. Linares-Vásquez, C. Mcmillan, D. Poshyvanyk, And M. Grechanik, “On Using Machine
Learning To Automatically Classify Software Applications into Domain Categories,” Empirical Software
Engineering, Vol. 19, No. 3, Pp. 582–618, 2014.
[7] S. Kawaguchi, P. K. Garg, M. Matsushita, And K. Inoue, “Mudablue: An Automatic
Categorization System For Open Source Repositories,” Journal Of Systems And Software, Vol. 79, No. 7,
Pp. 939–953, 2006.
[8] C. Mcmillan, M. Linares-Vasquez, D. Poshyvanyk, And M. Grechanik, “Categorizing Software
Applications For Maintenance,” in Software Maintenance (ICSM), 2011 27th IEEE International
Conference On. IEEE, 2011, Pp. 343–352.
[9] S. Vargas-Baldrich, M. Linares-Vásquez, And D. Poshyvanyk, “Automated Tagging of
Software Projects Using Bytecode And Dependencies (N),” in Automated Software Engineering (ASE),
2015 30th IEEE/ACM International Conference On. IEEE, 2015, Pp. 289–294.
[10] T. Wang, H. Wang, G. Yin, C. X. Ling, X. Li, And P. Zou, “Mining Software Profile Across
Multiple Repositories For Hierarchical Categorization,” in Software Maintenance (ICSM), 2013 29th IEEE
International Conference On. IEEE, 2013, Pp. 240–249.
[11] S. Ugurel, R. Krovetz, And C. L. Giles, “What’s The Code?: Automatic Classification of
Source Code Archives,” in Proceedings Of The Eighth ACM SIGKDD İnternational Conference on
Knowledge Discovery And Data Mining. ACM, 2002, Pp. 632–638.
[12] A. Kuhn, S. Ducasse, And T. Gírba, “Semantic Clustering: Identifying Topics in Source
Code,” Information And Software Technology, Vol. 49, No. 3, Pp. 230–243, 2007.
[13] P. S. Sandhu, J. Singh, And H. Singh, “Approaches For Categorization of Reusable Software
Components,” 2007.
58
[14] M. B. Kelly, J. S. Alexander, B. Adams, And A. E. Hassan, “Recovering A Balanced Overview
Of Topics in A Software Domain,” İn Source Code Analysis And Manipulation (SCAM), 2011 11th IEEE
International Working Conference On. IEEE, 2011, Pp. 135–144.
[15] A. Hindle, D. M. German, M. W. Godfrey, And R. C. Holt, “Automatic Classication of Large
Changes into Maintenance Categories,” İn Program Comprehension, 2009. ICPC’09. IEEE 17th
International Conference On. IEEE, 2009, Pp. 30–39.
[16] J. Escobar-Avila, M. Linares-Vásquez, And S. Haiduc, “Unsupervised Software
Categorization Using Bytecode,” in Proceedings Of The 2015 IEEE 23rd International Conference on
Program Comprehension. IEEE Press, 2015, Pp. 229–239.
[17] A. A. Al-Subaihin, F. Sarro, S. Black, L. Capra, M. Harman, Y. Jia, And Y. Zhang, “Clustering
Mobile Apps Based On Mined Textual Features,” İn Proceedings Of The 10th ACM/IEEE International
Symposium on Empirical Software Engineering And Measurement. ACM, 2016, P. 38.
[18] W. Martin, F. Sarro, Y. Jia, Y. Zhang, And M. Harman, “A Survey of App Store Analysis For
Software Engineering,” IEEE Transactions On Software Engineering, Vol. 43, No. 9, Pp. 817–847, 2017.
[19] A. T. Nguyen And T. N. Nguyen, “Automatic Categorization With Deep Neural Network For
Open-Source Java Projects,” in Proceedings Of The 39th International Conference On Software
Engineering Companion. IEEE Press, 2017, Pp. 164–166.
[20] A. Sharma, F. Thung, P. S. Kochhar, A. Sulistya, And D. Lo, “Cataloging Github
Repositories,” 2017.
[21] W. Chen, P. Xu, G. Wu, W. Dou, C. Gao, And J. Wei, “A Hierarchical Categorization
Approach For System Operation Services,” in Web Services (ICWS), 2017 IEEE Int’l Conf. On. IEEE,
2017, Pp. 700–707.
59
A New Nonparametric Test For Testing Equality of Locations Against Umbrella
Alternatives
Bülent ALTUNKAYNAK, Hamza GAMGAM, Merve BAĞÇACI
Department of Statistics, Gazi University, TURKEY
Abstract: In this study, a nonparametric new test is proposed to test the hypothesis of equality of
locations against umbrella alternatives. The Shan test for ordered alternatives is adapted to the
umbrella alternatives. This test can be considered as an extension of the sign test and the Wilcoxon
signed rank test. By a comprehensive simulation study, the proposed test is compared with the
Mack-Wolfe and Hettmansperger and Norton tests in terms of type I error rate and power. The
simulation results showed that all tests ensured the Bradley's robustness criteria for type I error
rate. The power comparison results indicated that the proposed test gives better results than the
other tests.
Keywords: Umbrella Alternative, Nonparametric Test, Monte Carlo, Simulation
REFERENCES
[1] J. T. Terpstra, "The Asymptotic Normality And Consistency of Kendall's Test Against Trend,
When Ties Are Present İn One Ranking," (İn En), Indagationes Mathematicae, Vol. 14, No. 3, Pp. 327333, 1952-01-01 1952.
[2] A. R. Jonckheere, "A Distribution-Free K-Sample Test Against Ordered Alternatives,"
Biometrika, Vol. 41, No. 1/2, Pp. 133-145, 1954.
[3] V. J. Chacko, "Testing Homogeneity Against Ordered Alternatives," (in En), Ann. Math.
Statist., Vol. 34, No. 3, Pp. 945-956, 1963/09 1963.
[4] M. L. Puri, "Some Distribution-Free K-Sample Rank Tests of Homogeneity Against Ordered
Alternatives," Communications On Pure And Applied Mathematics, Vol. 18, No. 1‐2, Pp. 51-63, 1965.
[5] R. B. May And P. R. Konkin, "A Nonparametric Test of An Ordered Hypothesis For K
Independent Samples," Vol. 30, No. 2, Pp. 251-257, 1970.
60
[6] R. E. Odeh, "On Jonckheere's K-Sample Test Against Ordered Alternatives," Technometrics,
Vol. 13, No. 4, Pp. 912-918, 1971/11/01 1971.
[7] J. Cuzick, "A Wilcoxon-Type Test For Trend," Vol. 4, No. 1, Pp. 87-90, 1985.
[8] T. P. Hettmansperger And R. M. Norton, "Tests For Patterned Alternatives in K-Sample
Problems," Journal Of The American Statistical Association, Vol. 82, No. 397, Pp. 292-299, 1987/03/01
1987.
[9] F. Beier And H. Buning, "An Adaptive Test Against Ordered Alternatives," (in English),
Computational Statistics & Data Analysis, Vol. 25, No. 4, Pp. 441-452, Sep 30 1997.
[10] M. Neuhäuser, P.-Y. Liu, And L. A. Hothorn, "Nonparametric Tests For Trend: Jonckheere's
Test, A Modification And A Maximum Test," Biometrical Journal, Vol. 40, No. 8, Pp. 899-909, 1998.
[11] H. Buning And W. Kossler, "Adaptive Tests For Umbrella Alternatives," (in English),
Biometrical Journal, Vol. 40, No. 5, Pp. 573-587, 1998.
[12] J. T. Terpstra And R. C. Magel, "A New Nonparametric Test For The Ordered Alternative
Problem," (in English), Journal Of Nonparametric Statistics, Vol. 15, No. 3, Pp. 289-301, Jun 2003.
[13] C. H. Chang And C. H. Yen, "A Nonparametric Test For The Ordered Alternative Based On
Fast Discrete Fourier Transform Coefficient," (in English), Journal of Testing And Evaluation, Vol. 39, No.
6, Pp. 1131-1143, Nov 2011.
[14] J. T. Terpstra, C. H. Chang, And R. C. Magel, "On The Use of Spearman's Correlation
Coefficient For Testing Ordered Alternatives," (in English), Journal Of Statistical Computation And
Simulation, Vol. 81, No. 11, Pp. 1381-1392, 2011.
[15] C. H. Chang, C. C. Chin, W. W. Yu, And Y. Y. Huang, "Using The Bernoulli Trial Approaches
For Detecting Ordered Alternatives," (in English), Bmc Medical Research Methodology, Vol. 13, Dec 5
2013.
[16] G. G. Shan, D. Young, And L. Kang, "A New Powerful Nonparametric Rank Test For Ordered
Alternative Problem," (in English), Plos One, Vol. 9, No. 11, Nov 18 2014.
[17] A. Gaur, "A Class of K-Sample Distribution-Free Tests For Location Against Ordered
Alternatives," (in English), Communications in Statistics-Theory And Methods, Vol. 46, No. 5, Pp. 23432353, 2017.
[18] G. A. Mack And D. A. Wolfe, "K-Sample Rank-Tests For Umbrella Alternatives," (in
English), Journal of The American Statistical Association, Vol. 76, No. 373, Pp. 175-181, 1981.
61
[19] W. Kossler, "Some C-Sample Rank Tests Of Homogeneity Against Umbrella Alternatives
With Unknown Peak," (in English), Journal Of Statistical Computation And Simulation, Vol. 76, No. 1,
Pp. 57-74, Jan 2006.
[20] A. Gaur, K. K. Mahajan, And S. Arora, "New Nonparametric Tests For Testing Homogeneity
of Scale Parameters Against Umbrella Alternative," (in English), Statistics & Probability Letters, Vol. 82,
No. 9, Pp. 1681- 1689, Sep 2012.
[21] N. Z. Shi, "Rank Test Statistics For Umbrella Alternatives," (in English), Communications in
Statistics-Theory And Methods, Vol. 17, No. 6, Pp. 2059-2073, 1988.
[22] A. Cohen And H. B. Sackrowitz, "Tests For The Umbrella Alternative Under Normality," (in
English), Communications İn Statistics-Theory And Methods, Vol. 25, No. 11, Pp. 2807-2817, 1996.
[23] G. H. Pan, "Distribution-Free Tests For Umbrella Alternatives," (in English), Communications
in Statistics-Theory And Methods, Vol. 25, No. 12, Pp. 3185-3194, 1996.
[24] B. A. Hartlaub And D. A. Wolfe, "Distribution-Free Ranked-Set Sample Procedures For
Umbrella Alternatives in The M-Sample Setting," (in English), Environmental And Ecological Statistics,
Vol. 6, No. 2, Pp. 105-118, Jun 1999.
[25] A. J. Hayter And W. Liu, "A New Test Against An Umbrella Alternative and The Associated
Simultaneous Confidence İntervals," (İn English), Computational Statistics & Data Analysis, Vol. 30, No.
4, Pp. 393-401, Jun 28 1999.
[26] W. C. Lee And Y. I. Chen, "Weighted Kaplan-Meier Tests For Umbrella Alternatives," (in
English), Annals of The Institute Of Statistical Mathematics, Vol. 53, No. 4, Pp. 835-852, Dec 2001.
[27] R. C. Magel And L. Qin, "A Non-Parametric Test For Umbrella Alternatives Based on
Ranked-Set Sampling," (İn English), Journal Of Applied Statistics, Vol. 30, No. 8, Pp. 925-937, Oct 2003.
[28] P. Singh And W. Liu, "A Test Against An Umbrella Ordered Alternative," (in English),
Computational Statistics & Data Analysis, Vol. 51, No. 3, Pp. 1957-1964, Dec 1 2006.
[29] M. Alvo, "Nonparametric Tests Of Hypotheses For Umbrella Alternatives," (in English),
Canadian Journal Of Statistics-Revue Canadienne De Statistique, Vol. 36, No. 1, Pp. 143-156, March 2008.
[30] S. V. Bhat And A. B. Patil, "A Class of K-Sample Distribution-Free Tests For The Umbrella
Alternatives," (in English), International Journal Of Agricultural And Statistical Sciences, Vol. 4, No. 1,
Pp. 53-61, Jun 2008.
62
[31] D. Basso And L. Salmaso, "A Permutation Test For Umbrella Alternatives," (in English),
Statistics And Computing, Vol. 21, No. 1, Pp. 45-54, Jan 2011.
[32] B. Altunkaynak And H. Gamgam, "Comparing The Performance of Nonparametric Tests For
Equality Of Location Against Ordered Alternatives," Communications in Statistics - Simulation And
Computation, Pp. 1-22, 2019.
[33] J. V. Bradley, "Robustness?, " British Journal of Mathematical And Statistical Psychology,
Vol. 31, No. 2, Pp. 144-152, 1978.
63
Smart Agriculture Applications with IoT
Mervenur SAĞLAM, Turgut ÖZSEVEN
Department of Mechatronic Engineering, Department of Computer Engineering
Tokat Gaziosmanpasa University, TURKEY
Abstract: Smart agriculture is the correct and economical use of resources, product control, and
production to be more ergonomic in order to increase production efficiency in agriculture. The
concept of internet of things that are widely used in conjunction with Industry 4.0 has started to be
applied in agriculture. The main objective is to apply the automation systems that provide
communication among themselves to agriculture in order to increase production efficiency. In this
study, the current state of the smart agricultural systems with IoT has been investigated. As a result,
the risks that may occur in production can be predicted in a short time with smart agriculture. Thus,
proper and wasteful use of the resources required for agriculture is foreseen.
Keywords: Smart Agriculture, Industry 4.0, Internet of Things
Introduction
The Internet of Things (IoT) is the communication of the interconnected objects over the network
without the need for people through mechanical and digital machines. In 1991, a study was
conducted at Cambridge University to track how much coffee was left in the coffee machine
without going to the machine [1]. In this study, the photo of the coffee pod was automatically taken
three times per minute and these photographs were transferred to the computer. Thus, the amount
of coffee used and the amount of coffee that was decreasing was followed in real time. This study
can be considered as the first steps of IoT [1].
Kevin Ashton, co-founder of MIT Auto-ID Center, first spoke about IoT in 1999 in Procher
& Gamble (P & G). The first industrial revolution included mechanical production facilities
between the 18th and 19th centuries. After 1870, the mass production of electrical energy was
called the second industrial revolution, while the third industrial revolution was called the digital
64
revolution in 1970. The basic structure of the 4th industrial revolution has created a new concept
with the existence of the internet concept [2].
The Industrial 4.0 or 4th Industrial Revolution was first used in 2011 at the Hannover Fair
in Germany. Industry 4.0, by improving living standards, technology, science, industrial
automation to bring together all the advanced applications to provide communication with sensors
[3]. The historical development of industry is given in Figure 1.
Fig. 1. Historical Development of Industry [4]
The technological developments in Industry 4.0 are given in Figure 2.
Fig. 2. Industry [5]
65
When Figure 2 is examined, Industry 4.0 includes many current technologies from
autonomous systems to cyber security, from virtual reality to simulation. The interaction of objects
through the Internet contributes to the user's ability to intervene in the device, the system to be
transformed into a smart state, to reduce the cost, to minimize the amount of work to be done and
thus to increase the profit.
In IoT, sensors, observation devices, chips, remote control systems are defined as objects.
That is, if a device is an object, it is also assumed to be intelligent. Thus, the device has a unique
ID. The sensor data is analyzed and transmitted via IDs to the device via the network. Nowadays,
IoT is used in many fields such as smart home automation, smart cities, industrial controls, energy
efficiency, health services, military applications. IoT sectoral distribution is given in Figure 3.
Fig. 3. IoT Sectoral Distribution [6].
Smart Agriculture
With the development of technology; terms such as intelligent agriculture, precision agriculture,
digital agriculture, farm management software are used. The most well-known smart agriculture
is the system of harvesting information. Smart agriculture and IoT provide remote access to data
such as remote control of the system, monitoring of moisture content, harvest time.
Innovations in the 4th industrial revolution have become a turning point in technology.
Cloud technology in smart agriculture systems, unmanned aerial vehicles, humidity-temperaturepressure sensors that can record all the land allow remote control of the system. Thus, it is aimed
66
to profit from time and product efficiency. The negativity of traditional agriculture is eliminated
by this transformation in agricultural production [4]. If the objectives of smart agriculture are
examined; to reduce the consumption of chemical materials, to minimize the damage to the
environment, to be able to easily archive the registration information of the produced product, to
obtain a higher quality of the produced product and to obtain a high quantity of product, to
determine the amount of production and to make a regular production.
When we summarize a general definition of smart agriculture; It includes the intervention
that can be done by taking into consideration the criteria that will be necessary for the needs that
differ in terms of location and time in the field where the agricultural operator makes an
agricultural production. It is difficult for farmers to evaluate the changes in heterogeneous
agricultural lands and to take precautions beforehand, and in some cases it is impossible. However,
with the development of technology, it can provide quantitative solutions to such problems. In this
way, it provides the possibility to make agricultural areas more efficient with the exception of
traditional methods [7, 8]. With the internet of things, the climate conditions in smart greenhouses
can be controlled by sensors and can be recorded. Thus, it can be interfering with remote control
of data instead of manually interfering.
Smart Agriculture With Iot in Turkey
Agriculture is one of the key areas that contribute to economic growth for Turkey. It is necessary
to maintain this growth by increasing production, quality and yield in agriculture. The proper use
of technology in the right place reduces the risks of agricultural products for operators. Correct
use of technology brings together the idea of agriculture. By using the IoT and machines and
intelligent systems, the least energy efficiency, water, seed, fertilizer and other agricultural
products are minimized.
Even if the producer knows little about the soil structure of the field, it can produce
different amounts of product on the same soil. The requirements of the cultivated product are
applied in the same amount on all sides of the land. For this reason, with the changing and
developing technology, timely irrigation, fertilization, spraying system will facilitate the
production process and prevent environmental pollution. After mechanization era in Turkey,
devices can communicate with each other, the objects that can operate synchronously is set right
67
timing by reducing the workload. In the field of agriculture, IoT systems can be used to manage
the process from production, storage and distribution of the plant.
One of the most important factors in agriculture is meteorological results. In the air station
systems developed with IoT; wind speed control, wind direction, temperature, pressure and
humidity values are obtained through the sensors to obtain early intervention in situations where
the risk is reduced by reducing the risk. An exemplary model of the processes that may be involved
in smart agriculture is given in Figure 4.
Fig. 4. Smart Agriculture [10]
The emphasis on smart agriculture in Turkey is increasing every day. State aids and
platforms have been established for this purpose (Smart Agricultural Platform). In this way, it is
aimed to easily detect and analyze the developments in the field of intelligent agriculture. In this
respect, early detection of potential future problems can be obtained [7].
Turkey intelligent agricultural areas, planning and pre-investment period. Domestic
technology manufacturers need to be supported and supported with innovation systems. Together
with smart agriculture, the young population will be able to obtain products with easier methods
and thus to agriculture. The example of a remote-controlled smart greenhouse is given in Figure
5.
68
Fig. 5. Smart Greenhouse Remote Monitoring Systems [10]
In Turkey, private firms, state-supported institutions, and smart agriculture is continuing R
& D efforts by individual users [4].
Nowadays, agricultural machinery has begun to integrate with intelligent technologies.
From the harvesting of the harvesting tractor, everything from the harvesting to the product output
in the field is recorded and big data are obtained. These data can be analyzed, and prospective
estimations can be made. With the modernization of agricultural machinery, data such as GPS,
wireless communication and snapshot can be obtained easily. An example of a greenhouse imaging
system is given in Figure 6.
Fig. 6. Greenhouse Monitoring [11]
69
As can be seen from Figure 6, all information in the greenhouse is taken wirelessly and can be
followed instantly from different platforms.
Results
The agriculture in Turkey is one of the biggest economic livelihoods. The importance given to
agriculture in Turkey with Industry 4.0 is increasing every day.
Intelligence of everything makes the use of sensors attractive in agricultural areas. In order to
obtain efficient product in agriculture, ambient values are measured and measured with sensors.
These data reduce the risks of agricultural products.
Considering that environmental pollution and organic food production are important and that water
resources are still inadequate, sensitive agriculture production should be transformed into
sustainable agriculture. Turkey is also satisfying the necessary infrastructure in this area,
manufacturers work should be done to make them conscious. Conscious producers bring conscious
consumer understanding.
REFERENCES
[1] https://www.karel.com.tr/blog/nesnelerin-interneti-nasil-dogdu, 27.03.2019.
[2] Drath, R., & Horch, A. (2014). Industrie 4.0: Hit or hype? [Industry forum]. IEEE industrial
electronics magazine, 8(2), 56-58.
[3] https://iiot-world.com/connected-industry/ industrie-4-0-is-all-about- speed, 25.03.2019.
[4]https://www.researchgate.net/publication/326550785_ Endustri_ 40_Ile_ Akilli_ Tarima_Gecis,
25.03.2019.
[5] Ünal, İ., & Topakcı, M. (2013). Tarımsal Üretim Uygulamalarında Bulut Hesaplama (Cloud
Computing) Teknolojisi. Akademik Bilişim Konferansı-AB, 23-25.
[6] https://growthenabler.com/flipbook/pdf/IOT%20Report.pdf, 11.04.2019.
[7] http://www.akillitarim.org/tr/-atprapor - 24.03.2019.
[8] Topakcı, M. ve Ünal, İ. “Hassas Tarımda Değişken Oranlı Uygulamalar”, Tarımsal
Mekanizasyon 26. Ulusal Kongresi, 22-23 Eylül, Hatay, Türkiye, (2010).
[9] https://semiengineering.com/farming-goes-high-tech/, 01/04/2019
70
[10] https://www.postscapes.com/smart-greenhouses/, 02/04/2019
[11] https://www.caipos.com/news/irriwave/, 02/04/2019
71
Teknoloji Kabul Modeli Kullanarak Netflix Platformu Kullanma
Maksadının Belirleyicileri
Ufuk CEBECİ, Oğuzhan İNCE
Department of Industrial Engineering, Istanbul Technical University, TURKEY
Abstract: Amazon Prime, Hulu, Apple TV, Puhutv, BluTV, Turkcell TV Plus vb. gibi digital
platformlar insanlar üzerindeki etkisini artırmaya devam ediyor ve Netflix bunlardan birisidir.
Netflix insanların seyretmek istediği film ve dizileri belirleyip onlara en kısa yoldan ulaştırmayı
hedefleyen bir platformdur. Bu platform günümüzde çok popüler hale gelmiştir ve sinema
sektörüyle yarışır seviyede bulunmaktadır. Netflix, veri madenciliğini etkili bir şekilde kullanarak
insanların neyi izlemeyi sevdiğini bilmektedir. Ayrıca, Netflix aşırı ve gereksiz verilerden
kurtularak büyük verilerin değerli bilgilere dönüştürülürken daha net olmasını sağladı. Bu bilgiler
ışığında ülkeler için ayrı dizi ve filmlerin çekilmesine öncülük etmiş, izleyicilerin sevdiği aktör ve
yönetmenleri bir araya getirerek, çektiği filmlerin izlenilirliğini arttırmış, böylece de Netflix
markasını dünyaya tanıtarak izleyicileri kendine bağlamayı başarmıştır. Bizim bu çalışmadaki
amacımız Netflix kullanıcılarının davranışlarını incelemek ve açıklamaktır. Bunu sağlamak için
anket çalışması yapılmış olup, Netflix platformunun kullanma maksadının belirleyicilerini
araştırmak için teknoloji kabul modeli kullanılmıştır.
Keywords: Veri Madenciliği Uygulamaları, Teknoloji Kabul Modeli, Netflix Kullanıcılarının
Davranışları
REFERENCES
[1] Amin, M. S., Chiam, Y. K., & Varathan, K. D. (2019). Identification of Significant Features
And Data Mining Techniques in Predicting Heart Disease. Telematics And Informatics, 36, 82-93.
[2] Bailey, A. A., Pentina, I., Mishra, A. S., & Ben Mimoun, M. S. (2017). Mobile Payments
Adoption By US Consumers: An Extended TAM. International Journal Of Retail & Distribution
Management, 45(6), 626-640.
72
[3] Boonsiritomachai, W., & Pitchayadejanant, K. (2017). Determinants Affecting Mobile Banking
Adoption By Generation Y Based on The Unified Theory Of Acceptance And Use of Technology Model
Modified By The Technology Acceptance Model Concept. Kasetsart Journal of Social Sciences.
[4] ÇELİK, H., & İPÇİOĞLU, İ. (2006). Gönüllü Teknoloji Kabulü: İnternet Kullanimini
Benimseme Davranişi Üzerine Bir Araştırma. Hacettepe Üniversitesi İktisadi ve İdari Bilimler Fakültesi
Dergisi, 24(1), 111-159.
[5] Davis, F. D. (1989). Perceived Usefulness, Perceived Ease of Use, and User Acceptance of
Information Technology. MIS Quarterly, 319-340.
[6] Dutot, V., Bhatiasevi, V., & Bellallahom, N. (2019). Applying The Technology Acceptance
Model İn A Three-Countries Study of Smartwatch Adoption. The Journal Of High Technology
Management Research.
[7] Estriegana, R., Medina-Merodio, J. A., & Barchino, R. (2019). Student Acceptance of Virtual
Laboratory And Practical Work: An Extension of The Technology Acceptance Model. Computers &
Education, 135, 1-14.
[8] Kim, C., Mirusmonov, M., & Lee, I. (2010). An Empirical Examination of Factors Influencing
the Intention To Use Mobile Payment. Computers in Human Behavior, 26(3), 310-322.
[9] Netflix 2019. Https://Media.Netflix.Com/En/About-Netflix.
[10] Ozturk, A. B., Bilgihan, A., Nusair, K., & Okumus, F. (2016). What Keeps The Mobile Hotel
Booking Users Loyal? Investigating The Roles Of Self-Efficacy, Compatibility, Perceived Ease of Use,
And Perceived Convenience. International Journal of Information Management, 36(6), 1350-1359.
[11] Pavlou, P. A. (2003). Consumer Acceptance Of Electronic Commerce: Integrating Trust And
Risk With the Technology Acceptance Model. International Journal of Electronic Commerce, 7(3), 101134.
[12] Scherer, R., Siddiq, F., & Tondeur, J. (2019). The Technology Acceptance Model (Tam): A
Meta-Analytic Structural Equation.
[13] Subramaniam, M., Iyer, B., & Venkatraman, V. (2019). Competing in Digital Ecosystems.
Business Horizons, 62(1), 83-94.
[14] Verma, P., & Sinha, N. (2018). Integrating Perceived Economic Wellbeing to Technology
Acceptance Model: The Case of Mobile Based Agricultural Extension Service. Technological Forecasting
and Social Change, 126, 207-216.
73
[15] Verma, S., Bhattacharyya, S. S., & Kumar, S. (2018). An Extension of The Technology
Acceptance Model in The Big Data Analytics System Implementation Environment. Information
Processing & Management, 54(5), 791-806.
[16] Wallace, L. G., & Sheetz, S. D. (2014). The Adoption Of Software Measures: A Technology
Acceptance Model (TAM) Perspective. Information & Management, 51(2), 249-259.
[17] Wu, J. H., & Wang, S. C. (2005). What Drives Mobile Commerce? An Empirical Evaluation
of The Revised Technology Acceptance Model. Information & Management, 42(5), 719-729.
[18] Yang, Y., & Wang, X. (2019). Modeling The İntention To Use Machine Translation For
Student Translators: An Extension of Technology Acceptance Model. Computers & Education, 133, 116126.
[19] Yoo, Y., Henfridsson, O., & Lyytinen, K. (2010). Research Commentary—The New
Organizing Logic of Digital İnnovation: An Agenda For Information Systems Research. Information
Systems Research, 21(4), 724-735.
[20] Zainab, B., Awais Bhatti, M., & Alshagawi, M. (2017). Factors Affecting E-Training
Adoption: An Examination Of Perceived Cost, Computer Self-Efficacy and The Technology Acceptance
Model. Behaviour & Information Technology, 36(12), 1261-1273.
74
Meta-Heuristic Methods Used in Optimization of SVM Learning Parameters
Zübeyir ÖZKORUCU, Turgut ÖZSEVEN
Department of Mechatronic Engineering, Department of Computer Engineering
Tokat Gaziosmanpasa University, TURKEY
Abstract: Support vector machine is an effective machine learning method based on statistical
learning theory and used for classification problems. The optimization of the parameter is very
important in order to increase the classification accuracy. Meta-heuristic methods are one of the
main optimization approaches that can be applied in this context and have been used frequently
for parameter optimization in recent years. These methods are generally particle swarm
optimization, genetic algorithm, grid search method, differential evolution algorithm, ant colony
optimization. In this study, support vector machine parameter optimization studies between 20102019 were investigated. According to the results of these studies, it was observed that parameter
optimization through meta-heuristic methods significantly increased the rate of classification
accuracy of classifier and significantly reduced the workload.
Keywords: Support Vector Machines, Meta-Heuristic Methods, Parameter Optimization,
Classification Accuracy
Introduction
The classification process plays an important role in machine learning and data mining. One of the
machine learning algorithms is Support Vector Machine (SVM). Based on statistical learning
theory and structural risk minimization, SVM is an effective method for solving pattern
recognition and classification problems. The basic idea in SVM is to determine a hyperplane that
separates the data appropriately from each other by maximizing the spacing of the nearest vectors
in the data set (Fig. 1-left) [1]. Although the number of hyperplanes may be more than one for the
data set which can be classified as linear (Fig. 1-right), only one of them makes the maximum
distance between the two classes, which is the hyperplane that the SVM must find.
75
Fig. 1. Optimal Hyperplane (Left) and Possible Hyperplanes (right) [2]
The studies to get the most appropriate result from a system under certain constraints are
called optimization. In other words, it is the process of determining the most appropriate values of
the variables of the problem in order to minimize or maximize the target function. The first thing
to do in the optimization process is to determine the decision parameters. Then the limiting
functions that define the values that the parameters cannot take, a cost function to be minimized
depending on the parameters or a profit function to be maximized are defined [3].
Meta-heuristic methods are algorithms that are proposed by inspiration from nature that
produces effective and appropriate solutions when traditional optimization methods produce
unacceptable solutions. These methods examine the search area effectively and efficiently and
conduct a solution research in a stochastic way. They set out from the set of solutions created in
each iteration and produce new solutions. However, they don't guarantee the global optimum
solution [4-6].
In this study, the studies conducted between 2010-2019 for the optimization of SVM
parameters were examined and analyzed. Analysis results are described in Section 3.
Used Optimization Methods For SVM
Meta-heuristic algorithms have been rapidly progressing in the literature in the last 20 years
because of their successful results [7-8]. Genetic Algorithm (GA) [9], Particle Swarm
Optimization (PSO) [10], Bat Algorithm (BA) [11], Differential Evolution Algorithm (DEA) [12],
Simulated Annealing (SA) [13] and Ant Colony Algorithm [14] are the most popular metaheuristic algorithms. These methods have advantages and disadvantages compared to each other
in terms of calculation complexity, classification accuracy rate and working time. The distribution
76
of meta-heuristic methods used in the studies examined and analyzed in this study is given in Fig.
2.
Fig. 2. Meta-heuristic Methods Used in Studies Examined
GA, which was proposed by John Holland (1975), is one of the evolutionary approaches
used in problems that are difficult to solve with traditional methods [9]. This algorithm is based
on the principle of protecting the lives of good generations and the destruction of the bad
generations.
PSO, which was proposed R.C. Eberhart and J.Kennedy in 1995, is an optimization
algorithm based on the behavior of bird flocks trying to find food [16]. Each individual in the
swarm is called a particle and particles form population. Each of these individuals is the candidate
solution. In this method, each particle adjusts its position to the best position by taking advantage
of its previous experience and individual with best position in swarm. Particles usually have a
better position than their previous position in their next position.
PSO has a fast convergence rate, but it is very easy to drop to the local optimum and has
premature convergence problems. Therefore, it is difficult to find the best solution in the solution
space. To overcome this problem, a mutation process was added to the PSO. Particles can be
reinitialized and updated to set the population search area after each iteration thanks to mutation
operation. In addition, this process prevents premature convergence by changing the position and
velocity of the particles and provides population diversity [17].
77
DEA, which was proposed by Price and Storn in 1995, is a population-based algorithm
[18]. This method gives effective results especially in problems with continuous data and its
operators and operation are similar to GA [12][18][19][20]. However, it is tried to improve the
performance of the solution by changing the usage of operators in DEA. The crossover, selection
and mutation operators used in GA are also used in this method. However, unlike GA, every
operator is not sequentially applied to the whole population. Moreover, chromosomes are taken
into consideration in this method and a new individual is obtained by using three other randomly
selected chromosomes. The fitness values between the existing chromosome and the new
chromosome obtained is compared and better one is transferred to the next population.
Ant Colony Optimization (ACO), which is created by Marco Dorigo, is a method inspired
by the methods of finding the shortest path between nests and food sources of ant colonies [21][22].
Dorigo has applied this method, which is a population based approach, to the Travelling Salesman
Problem and observed that ACO is highly effective in solving this problem. This method is widely
used in the solution of today’s optimization problems [14][21].
SA is a probability-based method proposed by Kirkpatrick et al. This method is called
simulated annealing because it is based on annealing process of solids. The annealing is to cool
the material slowly until it is crystallized after being heated to a certain maximum degree. If the
cooling is done appropriately, the crystal structure is very regular and a super cage structure is
obtained. If the cooling process is carried out very quickly, irregularities and disturbances occur
in the crystal structure. Therefore, the cooling process is very important. This method is especially
used in optimization of combinatorial problems which cannot be shown with mathematical models
[23].
BA is a population-based optimization algorithm based on the direction and distance
determining behavior of the object by utilizing the echoing of sound called echolocation [24].
Echolocation is a type of sonar, which many bat species use to communicate, to move without
hitting obstacles by perceiving objects around them, to determine the distance between itself and
its hunts [24][25][26]. All living things emit signals at a given frequency [26]. The bats listen to
the echo that occurs after they emit the signal and they determine the position of the objects and
prey around them by analyzing these echoes [26][27].
Artificial Chemical Reaction Optimization Algorithm (ACROA) is a probabilistic
optimization algorithm based on chemical reactions. During the chemical reaction, while bond
rupturing between some substances occurs, new bonds are formed between certain substances and
78
chemical changes occur. Thus, the energy and structure of reactants change. Furthermore, new
molecules formed as a result of the reaction can be used as reactants in another reaction. The new
molecules formed can be converted into their first reactants in two-way reactions. ACROA is a
method which is developed according to the types of chemical reactions and requires less
parameters [28][29][30]. Entropy for a maximization problem and enthalpy for a minimization
problem can be used as objective function in this method [28][30].
Elephant Herd Optimization (EHO) is one of the swarm intelligence algorithms proposed
by Wang, Den, Gao and Coelho in 2016. This method is based on the basic herd behaviors of
elephant groups. The elephant population consists of clans and each clan has a fixed number of
elephants. The elephants in each clan continue to live under the leadership of a matriarch. In every
generation, male elephants leave their clans and live away from them [31][32]. When we associate
these behaviors of the elephant herd with the optimization method, each elephant represents a
candidate solution. The population is divided into subgroups and forms the clans and the elephant
with the best fitness value in each clan is called the matriarch. After each iteration in the algorithm,
the worst elephant in the clan leaves from the herd and identifies a new location for itself. Each
elephant moves according to the clan's matriarch [32].
Grid search (GS) is a traditional method for parameter optimization. This method tries to
find all combinations with brute force and makes a comprehensive search. GS requires creating
two sets called learning rate and number of layers. The method trains the algorithm for all
combinations using these two sets and uses the Cross Validation (CV) method to measure
performance. Although GS is a simpler algorithm compared to other optimization algorithms, the
runtime is quite long, especially when the data set is too large [33].
Results of Existing Studies For SVM Optimization
According to the results of the studies examined within this study, the optimization of the
parameters in SVM has different effects on the results. This difference varies depending on the
optimized parameter and the method used.
Each kernel function has different kernel parameters in SVM. The optimization of learning
parameters in SVM is very important. Kernel parameters are learning parameters, too. One of
learning parameters is the penalty coefficient (С) that determines the trade-off between
maximization between class distances and the minimization of fault tolerance. When the value С
is greater, SVM results in fewer training errors and narrower distances, whereas it leads to more
79
distance and more educational errors in the small values [34]. However, the education data is also
very important in the high or low education error rate. This problem cannot be solved by only the
large value of С. Therefore, it is important to find an appropriate С value for the sample data in
the problem. Thus, the training model can establish a balance between sample data and educational
error. Another parameter is the kernel parameter that determines the complexity of the sample data
[35]. This parameter can both reduce complexity and guarantee classification accuracy. In short,
learning parameters have a great effect on the generalization ability and efficiency of SVM [36].
The optimization methods used in the studies examined in this article optimize these learning
parameters.
As a result of the studies examined, GA is a more suitable method than GS in optimization
of SVM parameters [37-39]. GS method results in a higher classification accuracy than original
SVM model [40], but when GA and GS are compared, GA increases the classification accuracy
more and the running time is shorter [37-39]. Although GA is a more favorable method than GS,
ACO [41], DE [42], ACROA [43], BA [44], EHO [37], M-PSO [45], PSO [46], SA [47] methods
are more successful than the GA method. DE [42], BA [44], M-PSO [45], ACROA [43] methods
are better than PSO in the studies examined. In the study, which has been used to optimize the
SVM parameters by using M-PSO [17], M-PSO method prevented falling to local optimum and
premature convergence problems of PSO.
Conclusion and Discussion
SVM is a powerful machine learning method used in classification problems. SVM parameter
optimization research has increased in recent years because of the heavy computational complexity
of SVM in training stage. Incorrect parameter selection in the SVM can affect the classification
performance of the method and increase the computational complexity [8]. Therefore, researchers
are looking for appropriate optimization algorithms and there are many studies in literature about
this topic [36-43]. In this article, the studies of SVM parameter optimization between 2010-2019
was analyzed. According to the results of the analysis,
Meta-heuristic methods used significantly increased the success of the classification
accuracy of SVM [17,35,37-40,42,44-46,48-57].
Most commonly used methods are GA and GS. However, ACO, SA, BA and PSO have
started to become widespread in recent years. Because these methods compared with GA
80
and GS in the studied examined, the rate of classification accuracy of these methods was
higher GA and GS [37,44].
Although meta-heuristic methods increase the accuracy of classification, these methods
take a long time to optimize the parameters in big data sets [58].
The adjustment of the kernel parameter and penalty coefficient in the SVM has vital
importance in increasing the classification accuracy.
The selection of the kernel function in increasing the classification success of SVM is also
very important.
New optimization algorithms are constantly introduced into the literature and these algorithms
are used in parameter optimization of SVM. In future studies, it is aimed to gain a new DVM
parameter optimization model by considering the success rates of the optimization algorithms used
in the studies examined.
REFERENCES
[1] V.N. Vapnik, The Nature of Statistical Learning Theory, 2nd ed., 2000, Springer-Verlag: New
York.
[2] T. Özseven, M. Ertürkler, M. Nurmuhammed, M. Gül and M. Harputluoğlu, “Quantifying the
Necrotic Areas on Liver Tissues Using Support Vector Machine (SVM) Algorithm and Gabor Filters,”
International Symposium on Innovations in Intelligent Systems and Applications Innovations in Intelligent
Systems and Applications, pp. 1-5, 2012.
[3] D. Karaboğa, Yapay Zekâ Optimizasyon Algoritmaları, 2nd ed., 2011, pp. 1-5.
[4] A. Onan, “Metaheuristic Methods and Their Application Areas,” Journal of Çukurova
University Faculty of Economics and Administrative Sciences, vol. 17,2013, pp.113-128.
[5] M. Yetkin, “Metasezgisel Algortimaların Jeodezi’de Kullanımı,” Journal Of Geomatic
Engineering Research, Vol. 1, 2016, Pp. 8-13.
[6] M. Yetkin, M. Berber, “Implementation of Robust Estimation İn GPS Networks Using The
Artificial Bee Colony Algorithm,” Earth Science Informatics, Vol. 7,2014, Pp. 39-46.
[7] X.-S. Yang, And I. Wiley, Engineering Optimization: An Introduction With Metaheuristic
Applications, 2010.
81
[8] B. Tezcan, “Determination Of Support Vector Machine Parameters By Meta Heuristic
Optimization Algorithms (MS Thesis)”, The Graduate School Of Natural And Applied Science Of Selçuk
University The Degree Of Master Of Science In Computer Engineering, 2018, Konya.
[9] J.H. Holland, Adaptation İn Natural And Artificial Systems. MIT Press, 1992.
[10] J. Kennedy And R. Eberhart, “Particle Swarm Optimization,” Proceedings Of ICNN'95 International Conference On Neural Networks, Vol. 1944, Pp. 1942-1948, Dec 1995.
[11] X.-S. Yang, “A New Metaheuristic Bat-Inspired Algorithm,” Nicso 2010: Nature Inspired
Cooperative Strategies For Optimization, Vol.284, 2010b, Pp.65-74.
[12] R. Storn, K. Price, “Differential Evolution - A Simple And Efficient Adaptive Scheme For
Global Optimization Over Continuous Spaces,” Journal Of Global Optimization, Vol. 23, 1995.
[13] S. Kirkpatrick, C.D Gelatt, M.P. Vecchi, “Optimization By Simulated Annealing,” Science,
Vol. 220, May 1983, Pp. 671-680.
[14] M. Dorigo, G.A. Di Caro, L.M. Gambardella, “Ant Algorithms For Discrete Optimization,”
Artificial Life, Vol. 5,1999, Pp. 137-172.
[15] M.N. Ab Wahab, S. Nefti-Meziani And A. Atyabi, “A Compherensive Review Of Swarm
Optimization Algorithms,” PLOS One, Vol. 23, 2015, Pp. 141-161.
[16] J. Kennedy And R. Eberhart, “Particle Swarm Optimization,” Proceedings Of ICNN'95 International Conference On Neural Networks, Vol. 1944, Pp. 1942-1948, Dec 1995.
[17] Z. He, S. Ding, B. Li, M. Yin And X. Zhang, “An İmproved Particle Swarm Optimization of
Support Vector Machine Parameters For Hyperspectral İmage Classification”, 12th IEEE Conference On
Industrial Electronics And Applications (ICIEA), Pp. 499-503, 2017.
[18] R. Storn, Http://Www1.İcsi.Berkeley.Edu/~Storn/Code.Html, [20.04.2019]
[19] D.G. Mayer, B.P. Kinghorn, A.A Archer, “Differential Evolution – An Easy And Efficient
Evolutionary Algorithm For Model Optimisation,” Agricultural Systems, Vol. 83, March 2005, Pp. 315328.
[20] T. Keskintürk, “Differential Evolution Algorithm,” Istanbul Commerce University Journal Of
Science, Vol. 5, 2006, Pp. 85-99.
[21] A.E. Şeneras, Ş. İnanç, “Ant Colony Optimization For Solving Vehicle Routing Problems,
Papers on Social Science, 2017, Pp. 58- 67.
82
[22] M. Demircioğlu, “A Heuristic Approach To Vehicle Routing Problem And An Application
(Doctoral Thesis),” Çukurova University Institute Business Administration Department, 2009, P. 163.
[23] S. Kirkpatrick, C.D Gelatt, M.P Vecchi, “Optimization By Simulated Annealing,” Science,
Vol. 220, May 1983 Pp. 671-680.
[24] L.L.K. Kamal, “Training Artificial Neural Networks Using Bat Optimization Algorithm (MS
Thesis),” The Graduate School Of Natural And Applied Science Of Selçuk University The Degree Of
Master Of Science In Computer Engineering, 2018, P. 62.
[25] M.B Fenton, “Bat Natural History And Echolocation,” Bat Echolocation Research: Tools,
Techniques And Analysis,2004, Pp. 2-6.
[26] S. Yılmaz, “Modification of Bat Algorithm on Unimodal, Multimodal and Shifted Numeric
Optimization Problems (MSc Thesis), “Süleyman Demirel University Graduate School of Applied And
Natural Sciences Department of Computer Engineering, 2014, P. 111.
[27] H.-U. Schnitzler, E.K. Kalko, “Echolocation By Insect-Eating Bats: We Define Four Distinct
Functional Groups Of Bats And Find Differences in Signal Structure That Correlate With The Typical
Echolocation Tasks Faced By Each Group,” Bioscience, Vol. 51, July 2001, Pp. 557-569.
[28] B. Alatas, A.B. Özer, “Mining Of Generalized Interesting Classification Rules With Artificial
Chemical Reaction Optimization Algorithm” Journal of The Faculty of Engineering And Architecture of
Gazi University, Vol. 32, 2017, Pp. 101-118.
[29] B. Alatas, “ACROA: Artificial Chemical Reaction Optimization Algorithm For Global
Optimization,” Expert Systems With Applications, Vol. 38, September 2011, Pp. 13170-13180.
[30] B. Alatas, “A Novel Chemistry Based Metaheuristic Optimization Method For Mining of
Classification Rules,” Expert Systems With Applications, Vol. 39, September 2012, Pp. 11080-11088.
[31] G.-G. Wang, S. Deb, X.-Z. Gao, and L. D. S. Coelho, “A New Metaheuristic Optimisation
Algorithm Motivated By Elephant Herding Behaviour,” International Journal of Bio-Inspired Computation,
vol. 8, 2016, pp. 394–409.
[32] H. Haklı, “Elephant Herding Optimization Using Multi-Search Strategy for Continuous
Optimization Problems,” Academic Platform Journal of Engineering and Science, vol. 7, 2019, pp. 261268.
[33] Prabhu, https://towardsdatascience.com/understandinghyperparameters- and-its-optimisation
-techniques-f0debba07568, 2018. [21.04.2019]
83
[34] T. Eitrich and B. Lang, “Efficient Optimization of Support Vector Machine Learning
Parameters For Unbalanced Datasets,” Journal of Computational and Applied Mathematics, vol. 196,
November 2006, pp. 425-436.
[35] J. Han, W. Jiang, C. Dai and H. Ma, “The Design of Diabetic Retinopathy Classifier Based on
Parameter Optimization SVM,” 2018 International Conference on Intelligent Informatics and Biomedical
Sciences (ICIIBMS), vol. 3, 2018, pp. 52-58.
[36] X. Zhang, X. Chen and Z. He, “An ACO-Based Algorithm For Parameter Optimization of
Support Vector Machines,” Expert Systems With Applications, vol.37, September 2010, pp. 6618-6628.
[37] E. Tuba, Z. Stanimirovic, “Elephant Herding Optimization Algorithm for Support Vector
Machine Parameters Tuning,” 9th International Conference on Electronics, Computers and Artificial
Intelligence (ECAI), pp. 1-4, 2017.
[38] I. Syarif, A. Prugel-Bennett, G. Wills, “SVM Parameter Optimization Using Grid Search and
Genetic Algorithm to Improve Classification Performance,” Telkomnika, vol. 14, 2016, pp. 1502- 1509.
[39] A.V. Phan, M.L. Nguyen and L.T. Bui, “Feature Weighting And SVM Parameters
Optimization Based on Genetic Algorithms For Classification Problems,” Applied Intelligence, vol. 46,
2017, pp. 455-469.
[40] L. Grama, L. Tuns and C. Rusu, “On the Optimization of SVM Kernel Parameters for
Improving Audio Classification Accuracy,” 14th International Conference on Engineering of Modern
Electric Systems (EMES), pp. 224-227, 2017.
[41] X. Zhang, X. Chen And Z. He, “An Aco-Based Algorithm For Parameter Optimization of
Support Vector Machines,” Expert Systems With Applications, Vol.37, September 2010, Pp. 6618-6628.
[42] L. Yang And Z. Xu, “Feature Extraction By Pca And Diagnosis of Breast Tumors Using Svm
With De-Based Parameter Tuning,” International Journal of Machine Learning And Cybernetics, Vol. 10,
2019, Pp. 591-601.
[43] H. Ao, J. Cheng, Y. Yang And T.K. Truong, “The Support Vector Machine Parameter
Optimization Method Based On Artificial Chemical Reaction Optimization Algorithm And İts Application
To Roller Bearing Fault Diagnosis,” Journal Of Vibration And Control, Vol. 21, 2015, Pp. 2434-2445.
[44] A. Tharwat, A.E. Hassanien And B.E. Elnaghi, “A Ba-Based Algorithm For Parameter
Optimization of Support Vector Machine,” Pattern Recognition Letters, Vol. 93, 2017, Pp. 13-22.
84
[45] C.-N. Chen, T.T. Hoang And M.-Y. Cho, “Parameter Optimisation Of Support Vector
Machine Using Mutant Particle Swarm Optimisation For Diagnosis of Metal-Oxide Surge Arrester
Conditions” Journal Of Experimental & Theoretical Artifical Intelligence, Vol. 31, 2019, Pp. 163-175.
[46] M. Hric, M. Chmulík And R. Jarina, “Model Parameters Selection For Svm Classification
Using Particle Swarm Optimization,” Proceedings of 21st International Conference Radioelektronika 2011,
Pp. 1-4, 2011.
[47] S. Wang, “Support Vector Machines Classification For Highdimentional Dataset,” Fourth
International Conference On Multimedia Information Networking And Security, Pp. 315-318, 2012.
[48] A. Özbeyaz, M.İ. Gürsoy, R. Çoban, “Regularization And Kernel Parameters Optimization
Based on Pso Algorithm in Eeg Signals Classification With SVM,” Ieee 19th Signal Processing And
Communications Applications Conference, Pp. 399-402, April 2011.
[49] L. Wang, G. Xu, J. Wang, S. Yang, L. Guo And W. Yan, “Ga-Svm Based Feature Selection
And Parameters Optimization For BCI Research,” Seventh International Conference on Natural Seventh
International Conference on Natural Computation, Vol. 1, Pp. 580- 583, 2011.
[50] L. Jun, D. Lixin And X. Ying, “Differential Evolution Based Parameters Selection For Support
Vector Machine,” Ninth International Conference on Computational Intelligence And Security, Pp. 284288, 2013.
[51] X.Z. Li And J.M. Kong, “Application Of Ga–Svm Method With Parameter Optimization For
Landslide Development Prediction,” Natural Hazards And Earth System Sciences Discussions, Vol. 14,
2014, Pp. 5295-5322.
[52] B.M. Sherin, And M.H. Supriya, “Ga Based Selection And Parameter Optimization For An
Svm Based Underwater Target Classifier,” International Symposium On Ocean Electronics (Sympol), Pp.
1-7, 2015.
[53] M.Yang, W. Li,H. Zhang And H. Wang, “Parameters Optimization Improvement of Svm On
Load Forecasting,” 8th International Conference On Intelligent Human-Machine Systems And Cybernetics
(Ihmsc), Vol. 02, Pp. 257-260, 2016.
[54] J. Manurung, H. Mawengkang And E. Zamzami, “Optimizing Support Vector Machine
Parameters With Genetic Algorithm For Credit Risk Assessment,” Journal Of Physics: Conference Series,
Vol. 930, 2017.
85
[55] M. -Y. Cho And T.T. Hoang, “Feature Selection And Parameters Optimization of Svm Using
Particle Swarm Optimization For Fault Classification İn Power Distribution Systems,” Computational
Intelligence And Neuroscience, Vol. 2017, 2017, P. 9.
[56] P. Tao, Z. Sun And Z.Sun, “An Improved Intrusion Detection Algorithm Based on Ga And
Svm,” Ieee Access, Vol. 6, 2018, Pp. 13624-13631.
[57] Z. Tao, L. Huiling, W. Wenwen And Y. Xia, “Ga-Svm Based Feature Selection and Parameter
Optimization İn Hospitalization Expense Modeling,” Applied Soft Computing, Vol. 75, February 2019, Pp.
323-332.
[58] Y. Liu, J. Du, “Parameter Optimization Of The Svm For Big Data,” 8th International
Symposium On Computational Intelligence And Design (Iscıd), Vol. 2, Pp. 341-344, 2015.
86
Determination Similarities of Basic Financial Indicators of Enterprises Included in the
NASDAQ Index Using by Hierarchical Clustering Distance Methods
Ebru Şeyma KARAKOYUN
Industrial Engineering Department, Karabük University, TURKEY
Abstract: Many indicators are used in the analysis of the financial status of companies. Indicators
are usually presented quarterly. However, some of the financial indicators of some companies are
not given data. In this study, it is examined whether there will be other indicators that Show the
same character instead of the data which is not present when the companies are evaluated as a
whole. A hierarchical clustering approach was used to evaluate whether the financial indicators of
10 randomly selected companies traded on the NASDAQ stock exchange to have similar
characteristics using distance methods. According to the results obtained, indicators with similar
characteristics are listed.
REFERENCES
[1] Rai, P., Singh, S. (2010). A Survey of Clustering Techniques, Int. J. Com. Appl. 7 (12) 1–5.
[2] Ando, T. And Bai J. (2016). Clustering Huge Number of Financial Time Series: A Panel Data
Approach With High-Dimensional Predictor And Factor Structures, Journal of The American Statistical
Association, İn Press.
[3] Kakizawa, Y., Shumway, R.H., Taniguchi, M. (1998). Discrimination And Clustering For
Multivariate Time Series, Journal of The American Statistical Association, 93:441, 328-340.
[4] Wang, X., Smith, K., Hyndman, R. (2006). Characteristic-Based Clustering For Time Series
Data, Data Min. Knowl. Discov. 13 (3) 335–364.
[5] Pavlidis, N., Plagianakos, V.P., Tasoulis, D.K., Vrahatis, M.N. (2006). Financial Forecasting
Through Unsupervised Clustering and Neural Networks, Oper. Res. 6 (2), 103–127.
[6] Scotto, M.G., Alonso, A.M., Barbosa, S.M. (2010). Clustering Time Series of Sea Levels:
Extreme Value Approach, J. Waterway Port, Coastal, Ocean Eng. 136 (4), 215–225.
87
[7] Gorji Sefidmazgi, M., Sayemuzzaman, M., Homaifar, A., Jha, M.K., Liess, S. (2014). Trend
Analysis Using Non-Stationary Time Series Clustering Based on The Finite Element Method, Nonlinear
Process. Geophys. 21 (3), 605–615.
[8] Ramoni, M., Sebastiani, P., Cohen, P. (2000). Multivariate Clustering By Dynamics, in:
Proceedings Of The National Conference On Artificial Intelligence, Pp. 633–638.
[9] Guo, C., Jia, H., Zhang, N. (2008). Time Series Clustering Based On ICA For Stock Data
Analysis, in: Proceedings Of 4th International Conference On Wireless Communications, Networking And
Mobile Computing, Wicom ’08, 2008, Pp. 1–4.
[10] Guan, H., Jiang, Q. (2007). Cluster Financial Time Series For Portfolio, in: ProceedingsoOf
The International Conference on Wavelet Analysis And Pattern Recognition, Pp. 851–856.
[11] Wang, X., Mueen, A., Ding, H., Trajcevski, G., Scheuermann, P., Keogh, E. (2012).
Experimental Comparison Of Representation Methods And Distance Measures For Time Series Data, Data
Min. Knowl. Discov., P. Springer Netherlands.
[12] Lkhagva, B., Suzuki, Y.U., Kawagoe, K. (2006). New Time Series Data Representation ESAX
For Financial Applications, in: Proceedings of 22nd International Conference on Data Engineering
Workshops, Pp. 17–22.
[13] Petitjean, F., Ketterlin, A., Gançarski, P. (2011). A Global Averaging Method For Dynamic
Time Warping, With Applications To Clustering, Pattern Recognit 44 (3), 678–693.
[14] Kakizawa, Y., Shumway, R.H., Taniguchi, M. (1998). Discrimination And Clustering For
Multivariate Time Series, J. Am. Stat. Assoc 93 (441), 328–340.
[15] Focardi, S. M. (2001). Clustering Economic And Financial Time Series: Exploring The
Existence of Stable Correlation Conditions. Technical Report 2001-04, The Intertek Group.
[16] Bao, D. (2007). A Generalized Model For Financial Time Series Representation and
Prediction, Appl. Intell. 29 (1), 1–11.
[17] Fu, T.C., Chung, F.L., Luk, R., Ng, C.M. (2010). Financial Time Series Indexing Based on
Low Resolution Clustering, in: Proceedings of The 4th IEEE International Conference on Data Mining
(ICDM-2004), Pp. 5–14.
[18] Ghassempour, S., Girosi, F., Maeder, A. (2014). Clustering Multivariate Time Series Using
Hidden Markov Models, Int. J. Environ. Res. Public Health 11 (3), 2741–2763.
88
[19] Spiegel S., Gaebler J., Lommatzsch A., De Luca E., Albayrak S., (2011). Pattern Recognition
And Classification For Multivariate Time Series, Sensorkdd ’11.
[20] An S., Gao X., Jiang M., Sun X., (2018). Multivariate Financial Time Series in The Light Of
Complex Network Analysis, Physica A 503, 1241–1255, 2018.
[21] Agrawal, R., Faloutsos, C., Swami, A. (1993). Efficient Similarity Search in Sequence
Databases, Found. Data Organ. Algorithms 46, 69–84.
[22] Keogh, E. (2002). Exact İndexing Of Dynamic Time Warping. In 28th International
Conference on Very Large Databases, Hong Kong. Pp. 406-417.
[23] Berndt, D. & Clifford, J. (1994). Using Dynamic Time Warping To Find Patterns in Time
Series. AAAI-94 Workshop On Knowledge Discovery İn Databases.
[24] Eiter, T., & Mannila, H. (1994). Computing Discrete Fréchet Distance. Tech. Report CD-TR
94/64, Information Systems Department, Technical University Of Vienna.
[25] Drimel A., Krivošija A., Sohler C., (2015). Clustering Time Series Under The Frèchet
Distance, Arxiv.Org.
[26] Casado De Lucas D (2010). Classification Techniques For Time Series and Functional Data.
Ph.D. Thesis, Universidad Carlos III De Madrid.
[27] Keogh, E. And Kasetty, S., (2002). On The Need For Time Series Data Mining Benchmarks:
A Survey And Empirical Demonstration. In Proceedings Of The 8th ACM SIGKDD Int'l Conference On
Knowledge Discovery And Data Mining. Edmonton, Alberta, Canada, July 23-26. Pp. 102-111.
89
RFM Model for Segmentation in Retail Analytics: A Case Study
İnanç KABASAKAL
Faculty of Economics and Administrative Sciences, Ege University, TURKEY
Abstract: Over the last decades, marketing scholars have often drawn attention to the value of
customers for businesses that aim to endure in a harsh competitive environment. Customer
Relationship Management (CRM) has been a prominent approach in marketing management that
aims to improve relationships with customers. A practical implication of the CRM approach is the
analysis of customer data to extract value for businesses, as well as customers. Segmentation has
been a useful task that helps to group customers with similar attributes and designate better-tailored
marketing strategies for customer groups. Among a variety of approaches for customer
segmentation, Recency Frequency Monetary (RFM) Model stands out as an easy-to-adopt and
effective technique. In this study, segmentation with RFM approach will be conducted over the
purchase records obtained from an e-retailer. The segments and relevant marketing strategies will
be presented in the findings. Moreover, a software implementation for the RFM model will be
introduced along with a case study.
Keywords: Business Analytics, Analytical CRM, Customer Segmentation, RFM Model
REFERENCES
[1] E. Chablo, “The İmportance Of Marketing Data İntelligence in Delivering Successful CRM”,
in Customer Relationship Management, Vieweg+ Teubner Verlag, Wiesbaden, 2000, Pp. 57–70.
[2] S. Gupta, D. R. Lehmann. “Customers As Assets”. Journal Of Interactive Marketing, 2003,
17(1), 9–24.
[3] P. E. Pfeifer. “The Optimal Ratio Of Acquisition And Retention Costs”, Journal of Targeting,
Measurement And Analysis For Marketing, 2005, 13(2), 179–188.
[4] Han, J. and Kamber, M. (2006) Data Mining Concepts and Techniques. 2nd Edition, Morgan
Kaufmann Publishers, San Francisco.
90
[5] C. H. Cheng, And Y.S. Chen. “Classifying The Segmentation Of Customer Value Via Rfm
Model And Rs Theory”, Expert Systems With Applications, 2009, 36(3), Pp. 4176–4184.
[6] K. Elliott, R. Scionti, And M. Page. “The Confluence Of Data Mining And Market Research
For Smarter Crm”, Spss White Paper, Chicago, 2003, Pp. 1–11.
[7] M. Xu, And J. Walton. “Gaining Customer Knowledge Through Analytical CRM”, Industrial
Management & Data Systems, 2005, 105(7), Pp. 955–971.
[8] S.A.C. Madeira. “Comparison Of Target Selection Methods in Direct Marketing”, 2002,
Unpublished Msc. Thesis.
[9] J. T. Wei, S. Y. Lin, And H. H. Wu. “A Review of The Application Of RFM Model”, African
Journal of Business Management, 2010, 4(19), 4199–4206.
[10] V. Aggelis, And D. Christodoulakis. “Customer Clustering Using RFM Analysis”, in
Proceedings of The 9th World Scientific and Engineering Academy And Society International Conference
on Computers, July 2005, Pp. 2–7.
[11] Kumar, V. “Clv: The Databased Approach”, Journal of Relationship Marketing, 2006, 5(2-3),
7–35.
[12] M. Khajvand, K. Zolfaghar, S. Ashoori, And S. Alizadeh. “Estimating Customer Lifetime
Value Based on Rfm Analysis of Customer Purchase Behavior: Case Study”, Procedia Computer Science,
2011, 3, 57–63.
[13] B. Sohrabi, A. Khanlari. “Customer Lifetime Value (Clv) Measurement Based on Rfm
Model”, Iranian Accounting & Auditing Review, 2007, 14(47), Pp. 7–20.
[14] Sqlite Official Website. “About Sqlite”, Url: Https://Www.Sqlite.Org/Index.Html, Date
Accessed: 25.04.2019.
[15] Sqlite Official Website. “Sqlite: Most Widely Deployed and Used Database Engine”, Url:
Https://Sqlite.Org/Mostdeployed.Html, Date Accessed: 25.04.2019.
[16] M. Fuchs, W. Höpken, And M. Lexhagen. “Big Data Analytics For Knowledge Generation in
Tourism Destinations – A Case From Sweden”, Journal Of Destination Marketing & Management, 2014,
3(4), Pp.198–209.
[17]
S.
Sayad.
“An
Introduction
to
Data
Science:
Https://Www.Saedsayad.Com/Data_Preparation.Htm, Date Accessed: 20.04.2019.
91
Data
Preparation”,
[18] K. K. Tsiptsis, And A. Chorianopoulos. “Data Mining Techniques in Crm: Inside Customer
Segmentation”, John Wiley & Sons, 2009.
[19]
Putler
Blog.
“Rfm
Analysis
For
Successful
Https://Www.Putler.Com/Rfm-Analysis, Access Date: 26.04.2019.
92
Customer
Segmentation”,
Url:
Siber Tehdit İstihbaratı Alanında Makine Öğrenmesi Algoritmalarının Kullanılması
Cemile SARICAOĞLU, Mehmet DEMİRCİ
Department of Computer Engineering, Gazi University, TURKEY
Abstract: Nowadays, with the developing technology, the amount of data that is owned and
processed is increasing day by day. It is very important to ensure the security of data, which is one
of the biggest assets for institutions and organizations. With traditional security methods, attacks
can be detected and prevented, but cybercriminals spend a lot of time and resources on advanced
and targeted attacks that can bypass these methods. The present methods are reactive because they
are generally updated with the information obtained from the analyzes performed after a successful
attack. More proactive approaches are needed to improve safety. Cyber threat intelligence
represents such a proactive approach and involves collecting and analyzing information for
potential threats from a wide variety of data sources. The purpose of cyber-threat intelligence is to
proactively adapt security controls to understand the methodology used by different attackers and
to detect and prevent such activities. In the world of technology, the defense against attacks is one
of the most important issues. Today, different approaches and effective methods have been used
to obtain intelligence. These include vital information about security threats, which are used by
hacker forums and other platforms as a means of communication between hackers. The amount of
data on such platforms is very large. The manual analysis of these data is time-consuming,
ineffective and requires a considerable amount of resources. In this sense, machine learning has
become one of the popular approaches used in the field of cyber-threat intelligence in terms of its
suitability to the subject, producing beneficial and effective results. In this study, information is
given about cyber threat intelligence and in the world of hackers, how to obtain intelligence by
using machine learning techniques is examined and evaluated in detail by supporting the studies
conducted in the literature.
Keywords: Cyber Security, Threat Intelligence, Machine Learning, Hacker Forums
93
REFERENCES
[1] Webroot Threat Intelligence: What is it, How Can it Protect You From Today's Advanced
Cyber-Attacks?URL:Https://Www.Gartner.Com/İmagesrv/Mediaproducts/Pdf/Webroot/İssue1_ Webroot.
Pdf
[2] Shackleford, D. The SANS State Of Cyber Threat Intelligence Survey: CTI Important And
Maturing.
Technical
Report,
SANS
Institute,
2016.
URL:
Https://Www.Sans.Org/Reading-
Room/Whitepapers/Bestprac/Statecyber-Threat-İntelligence-Survey-Cti-İmportant-Maturing-37177.
[3] Türkmenoğlu, C. (2016). Türkçe Metinlerde Duygu Analizi (Doctoral Dissertation, Fen
Bilimleri Enstitüsü).
[4] Alpaydin, E. (2004). Introduction to Machine Learning (Adaptive Computation And Machine
Learning Series). The MIT Press Cambridge.
[5] Öztemel E., “Yapay Sinir Ağları”, Papatya Yayıncılık, İstanbul (2003).
[6] Koyuncugil, A., & Özgülbaş, N. (2009). Veri Madenciliği: Tıp ve Sağlık Hizmetlerinde
Kullanımı ve Uygulamaları. Bilişim Teknolojileri Dergisi, 2(2).
[7] Çoban, O. (2016). Metin Sınıflandırma Teknikleri ile Türkçe Twitter Duygu Analizi. Yüksek
Lisans Tezi, Atatürk Üniversitesi, Fen Bilimleri Enstitüs.
[8] Ardıl, E. (2009). Esnek Hesaplama Yaklaşımı ile Yazılım Hata Kestrimi (Master's Thesis).
[9] KARTAL, E., Programı, E., & Balaban, M. E. Sınıflandırmaya Dayalı Makine Öğrenmesi
Teknikleri ve Kardiyolojik Risk Değerlendirmesine İlişkin Bir Uygulama.
[10] Shearer, C. (2000). The CRISP-DM Model: The New Blueprint For Data Mining. Journal of
Data Warehousing, 5(4), 13-22.
[11] Deliu, I. (2017). Extracting Cyber Threat İntelligence From Hacker Forums (Master's Thesis,
NTNU).
[12] Nunes, E., Diab, A., Gunn, A., Marin, E., Mishra, V., Paliath, V& Shakarian, P. (2016,
September). Darknet and Deepnet Mining For Proactive Cybersecurity Threat İntelligence. In 2016 IEEE
Conference on Intelligence And Security Informatics (ISI) (Pp. 7-12).
[13] Marin, E., Diab, A., & Shakarian, P. (2016, September). Product Offerings in Malicious
Hacker Markets. In 2016 IEEE Conference on İntelligence And Security İnformatics (ISI) (Pp. 187-189).
IEEE.
94
[14] Samtani, S., Chinn, R., & Chen, H. (2015, May). Exploring Hacker Assets in Underground
Forums. In 2015 IEEE International Conference on Intelligence and Security Informatics (ISI) (Pp. 31-36).
IEEE.
[15] Samtani, S., Chinn, K., Larson, C., & Chen, H. (2016, September). Azsecure Hacker Assets
Portal: Cyber Threat İntelligence And Malware Analysis. In 2016 IEEE Conference on Intelligence And
Security Informatics (ISI) (Pp. 19-24). Ieee.
[16] Fang, Z., Zhao, X., Wei, Q., Chen, G., Zhang, Y., Xing, C.& Chen, H. (2016, September).
Exploring Key Hackers And Cybersecurity Threats İn Chinese Hacker Communities. In 2016 IEEE
Conference On Intelligence And Security Informatics (ISI)(Pp. 13-18). IEEE.
[17] Benjamin, V., Li, W., Holt, T., & Chen, H. (2015, May). Exploring Threats And
Vulnerabilities in Hacker Web: Forums, IRC And Carding Shops. In 2015 IEEE International Conference
On Intelligence And Security Informatics (ISI) (Pp. 85-90). IEEE.
[18] Macdonald, M., Frank, R., Mei, J., & Monk, B. (2015, August). Identifying Digital Threats in
A Hacker Web Forum. In 2015 IEEE/ACM International Conference On Advances İn Social Networks
Analysis And Mining (ASONAM) (Pp. 926-933). IEEE.
95
Analysis of Non-Risked Provinces; Unemployment and Traffic Accidents
Taner ERSÖZ, Betül CEBESOY
Department of Actuary and Risk Management, Department of Industrial Engineering
Karabük University, TURKEY
Abstract: Risk is any possibility that affects the achievement of the intended objectives. In this
study, it was investigated which provinces did not contain any risks. In the calculation of risk
values of provinces; unemployment and traffic accidents were taken into consideration. Turkey
Statistical Institute (TURKSTAT), the General Directorate of Security, the Social Security
Institution (SSI) and Turkey Business Association (TBA) risk values obtained from benefiting
from the data of the years 2013 to 2017 were calculated. Risk values are evaluated between 1 and
5. The green-colored provinces are risk-free, while the red-colored region is cons idered to be of
high risk. Fine-Kinney method was used for risk analysis.
Keywords: Free Risk Zone, Risk Analysis, Fine-Kinney Method, Risk Map
96
Use of Grid Search in Hyper-Parameter Selection for Time Series Analysis: A Case Study
with Ad Mediation Software
Görkem GİRAY1, Murat Osman ÜNALIR2, Şeyma TAHMAZ1
1
2
Kokteyl A.Ş.
Computer Engineering Department, Ege University, TURKEY
Abstract: The success of an Ad Mediation Software’s decision on from which ad network to
request ads and in which order depends on the ability to estimate eCPM (effective Cost Per Mille)
value used to measure ad revenue. This value varies for different applications depending on
different external factors. It is not possible for domain experts to make successful predictions by
analyzing different sets of external factors for a large number of applications and to keep these
estimates constantly up to date. Therefore, eCPM values were automatically estimated separately
for each application based on different advertising spaces and different countries using time series
analysis. The ARIMA model was used to estimate and the hyper-parameters of the model were
optimized by using grid search method. For most of the values obtained, it was found that the
values obtained with the intuition of domain experts were closer to the actual values.
Keywords: Time Series Analysis, ARIMA, Hyper-parameter Optimization, Grid Search Method,
Ad Mediation Software
Giriş
Mobil cihazların ve bununla paralel olarak mobil uygulamaların kullanımı son yıllarda oldukça
yaygınlaşmıştır. Bu uygulamaların gelirlerinin önemli bir bölümü reklam gösteriminden
gelmektedir. Bunun sonucu olarak mobil uygulamalara reklam sağlayan çok sayıda reklam ağı
ortaya çıkmıştır. Reklam ağları, reklam verenlerden reklam alarak bu reklamları uygun yayıncılara
(bu bildiri kapsamında mobil uygulamalara) sağlamaktadır. Çok sayıda reklam ağının ortaya
çıkmasıyla birlikte bir yayıncının kendisine en çok kazanç sağlayacak reklamı yayımlayabilmesi
için çok sayıda reklam ağı ile çalışması gerekliliği ortaya çıkmıştır. Bu karmaşıklığı yayıncıların
97
yerine yönetmek için de reklam ağları ile reklam yayıncıları arasında reklam aracıları
konumlanmıştır.
Bir reklam aracısı, bir uygulamaya, yani yayıncıya, reklam sağlarken, hangi reklam
ağlarından hangi sırada reklam talep edeceğine dair bir karar verir. Bu kararı verirken uygulamanın
reklam gelirini (aynı zamanda da kendi reklam gelirini) azami seviyeye çıkarmayı hedefler. Bunun
için de reklam gösteriminden en fazla gelir sağlanabilecek reklam ağından başlayarak reklam
talebinde bulunur. Mobil reklam sektöründe reklam gösterimi başına geliri ifade etmek için
eBGBM (Etkin Bin Gösterim Başına Maliyet; İngilizcesi: eCPM – Effective Cost per Mille)
kavramı kullanılmaktadır. Mevcut durumda Kokteyl şirketinin Reklam Aracısı Yazılımında
(RAY) gerçekleşecek eBGBM değerleri alan uzmanları tarafından sezgiye dayalı olarak tahmin
edilmektedir. eBGBM değerini etkileyebilecek değişkenler tespit edilerek, bu değişkenlerle
eBGBM arasındaki örüntülerin makine öğrenmesi algoritmalarıyla tespit edilmesi mümkün
görünmektedir [1 - 4]. Bu çalışma kapsamında mevcut durumda sezgisel olarak yapılan
tahminlerin başarısı istatistiksel yöntemler kullanılarak incelenmiştir. Elde edilen sonuçlara göre
başka yöntemler kullanılarak daha iyi tahmin yapılabileceği görülmüştür.
Bildirinin geri kalan bölümleri şu şekilde düzenlenmiştir: İkinci bölüm mobil reklam
sektörü ve RAY hakkında genel bir bilgi vermektedir. Üçüncü bölümde problem tanımı, beşinci
bölümde ise bu problem için geliştirilen çözüm anlatılmaktadır. Son bölümde sonuçlar ve gelecek
çalışmalar sunulmuştur.
Arka Plan
Şekil 1 mobil reklam sektöründeki ana paydaşları göstermektedir. Reklam verenler, reklam
ağlarına ürünleri ve/veya hizmetleri hakkında reklam vermektedir. Yayıncılar, reklam ağlarından
reklam talep etmekte ve reklam ağları yayıncıya o andaki en uygun reklamı sağlamaktadır. Uygun
reklamın seçimi konusundaki karar birçok değişkene (yayıncının hangi uygulama olduğu, son
kullanıcı hakkındaki bilgiler, uygulamanın çalıştığı cihaz gibi) bağlı olarak değişmektedir.
Yayıncıların reklam ağlarından doğrudan reklam alması durumunda bazı zorluklar ortaya
çıkmaktadır. Uygulamanın geliştirilmesi açısından ortaya çıkan bir zorluk, yayıncı uygulamanın
reklam talep edeceği tüm reklam ağlarıyla ayrı ayrı entegre olma gereksinimidir. İş amaçları
açısından bakıldığında bir yayıncının birçok reklam ağından en uygun reklamı (en fazla geliri
getirecek reklam) alamaması durumunda gelir kaybı yaşanacaktır. Dolayısıyla yayıncının en
98
uygun reklamı seçecek karar algoritmasını uygulamanın bir parçası olarak geliştirmesi
gerekmektedir. Reklam aracıları birçok reklam ağı ile entegre olarak en uygun reklamı sağlama
konusunda uzmanlaşmaktadır. Uygulama geliştirme açısından yayıncı uygulamanın sadece bir
reklam aracısı ile entegre olarak birçok reklam ağından reklam alabilmektedir ve çok sayıda
reklam ağı ile entegre olmanın getirdiği karmaşıklık reklam aracısı tarafından yönetilmektedir.
Bunun yanında en uygun reklamı seçme konusunda uzmanlaşan reklam aracıları bu işi de yayıncı
uygulamaların yerine yapmaktadır. Böylece yayıncıların reklam gelirlerini arttırmasına yardımcı
olmaktadır. Son kullanıcılar ise yayıncı uygulamaları kullanarak reklamları görüntülemekte ve bir
davranış
(reklama
tıklama,
uygulama
indirme
ve
kurma,
ürün
sergileyebilmektedir. Bu davranışlar yayıncılara gelir sağlamaktadır.
Şekil 1. Mobil Reklam Sektöründeki Ana Paydaşlar.
99
satın
alma
gibi)
Şekil 2. Reklam Aracısı Yazılımının Yayıncılar Ve Reklam Ağları İle İlişkisi.
Şekil 2’de RAY’nin (bu yazılımın mimarisi [5] numaralı kaynakta anlatılmaktadır)
yayıncılarla ve reklam ağlarıyla olan ilişkisi gösterilmektedir. RAY’yi oluşturan iki bileşen yeşil
renk ile gösterilmiştir. RAY sunucu uygulaması, reklam ağlarının yazılım geliştirme kitlerini
(YGK) içermektedir. Bu YGK’ler (Şekil 2’deki Reklam Ağı 1’in ve 2’nin YGK’si) aracılığıyla
sunucu uygulaması reklam ağlarından reklam talep etmektedir ve reklam almaktadır. Yayıncı
uygulamalar ise RAY’nin YGK’sini uygulamalarının içine yerleştirerek RAY’den reklam talep
etmektedir ve almaktadır. Daha önce anlatıldığı gibi, teknik açıdan RAY birçok reklam ağı ile
entegre olmanın getirdiği karmaşıklığı yayıncı uygulamalara saydam hale getirmektedir. İş
amaçları açısından ise RAY birçok reklam ağından sağladığı reklamlar arasından en uygun reklamı
yayıncı uygulamaya göndererek reklam gelirlerinin arttırılmasını sağlamaktadır.
Problem Tanımı
Şekil 2’de gösterilen RAY’nin içerdiği bir bileşen olan şelale listesi üretim motorunun ürettiği
şelale listesi, yayıncının hangi reklam ağlarından hangi sırada reklam talebinde bulunması
100
gerektiğini belirleyen bir öncelik listesidir (Bkz. Şekil 3). Şelale listesi, her bir yayıncı için belirli
aralıklarla (örneğin günlük) çeşitli değişkenler (reklam formatı, reklam alanı, geçmiş eBGBM
değerleri gibi) göz önüne alınarak şelale listesi üretim motoru tarafından üretilir. Her yayıncı
uygulamanın RAY YGK’si bu şelale listesini belirli aralıklarla sunucu uygulamasından indirerek
reklam taleplerini bu güncellenmiş listeye göre yapar. Şelale listesindeki sıralama gelecekte
gerçekleşecek eBGBM değerlerini ne kadar iyi yansıtırsa reklam gelirleri o kadar artacaktır.
Şekil 3. Örnek Bir Şelale Listesi.
eBGBM reklam gösterimi başına geliri ifade etmektedir ve “(Gelir / Gösterim Sayısı) x
1000” formülü ile hesaplanmaktadır. Mevcut durumda şelale listesi oluşturmak için kullanılan
girdiler (eBGBM tahminleri gibi) alan uzmanları tarafından yapılan gözlemler sonucunda sezgisel
olarak belirlenmektedir ve güncellenmektedir.
Gerçekleşen eBGBM değerleri ve bağlantılı veriler RAY’de Tablo 1’de gösterilen
alanlarda saklanmaktadır. Her yayıncı uygulamanın her bir reklam alanı için ülke bazında günlük
olarak gerçekleşen ve daha önce tahmin edilen eBGBM değerleri RAY tarafından
kaydedilmektedir. Her ülkede mobil uygulamaların kullanım dinamiklerinin farklı olması
nedeniyle eBGBM değerleri ülke bazında ayrı ayrı tahmin edilmektedir. Bunun yanında reklam
ağları da her yayıncı uygulama için ülke bazında gerçekleşen eBGBM değerlerini reklam
aracılarıyla paylaşmaktadır.
Örnek olarak bir yayıncı uygulama için (uygulamanın adı gizlilik nedeniyle verilmemiştir)
üç günde, dört ülkede gerçekleşen ve daha önce tahmin edilen eBGBM değerleri Tablo 2’de
gösterilmektedir.
101
Tablo 1. eBGBM Değerlerinin ve Bağlantılı Verilerin Saklandığı Alanlar Ve Açıklamaları
Alan adı (Türkçe)
Tarih
Reklam alanı no
Alan adı (İngilizce)
date
placement_id
Yayıncı uygulamanın adı
Ülke
Gerçekleşen eBGBM
Tahmin edilen eBGBM
app_name
country
real_ecpm
weighted_ecpm
Açıklama
tarih
yayıncı uygulamanın reklam gösterilebilecek bir
konumunun RAY’deki kimlik numarası
yayıncı uygulamanın adı
yayıncı uygulamanın kullanıldığı ülke
gerçekleşen eBGBM değeri
alan uzmanları tarafından tahmin edilen eBGBM
değeri
Tablo 2. Örnek eBGBM Değerleri ve Bağlantılı Veriler
Tarih
Reklam alanı no
08-09-18
00a78a52-7572-4ab6-863c-df5eb77ebe99
09-09-18
00a78a52-7572-4ab6-863c-df5eb77ebe99
10-09-18
00a78a52-7572-4ab6-863c-df5eb77ebe99
08-09-18
00a78a52-7572-4ab6-863c-df5eb77ebe99
09-09-18
00a78a52-7572-4ab6-863c-df5eb77ebe99
10-09-18
00a78a52-7572-4ab6-863c-df5eb77ebe99
08-09-18
00a78a52-7572-4ab6-863c-df5eb77ebe99
09-09-18
00a78a52-7572-4ab6-863c-df5eb77ebe99
10-09-18
00a78a52-7572-4ab6-863c-df5eb77ebe99
08-09-18
00a78a52-7572-4ab6-863c-df5eb77ebe99
09-09-18
00a78a52-7572-4ab6-863c-df5eb77ebe99
10-09-18
00a78a52-7572-4ab6-863c-df5eb77ebe99
Uygulamanın
adı
yayıncı
uygulama
yayıncı
uygulama
yayıncı
uygulama
yayıncı
uygulama
yayıncı
uygulama
yayıncı
uygulama
yayıncı
uygulama
yayıncı
uygulama
yayıncı
uygulama
yayıncı
uygulama
yayıncı
uygulama
yayıncı
uygulama
Gerçekleşen
eBGBM
Tahmin
edilen
eBGBM
1.280701754
0.012145187
BR
0.482967144
0.240236608
BR
0.769817073
0.323807174
BR
5.929345644
0.0260193
DE
1.026119403
0.429765486
DE
1.94153337
0.460431981
DE
6.718092567
0.021978022
FR
0.885636856
0.549004147
FR
2.264134582
0.626127083
FR
1.640059765
0.005023753
TR
0.515668347
0.561521377
TR
0.684581689
0.535929221
TR
Ülke
Çözüm Önerisi
Alan uzmanlarının sezgileriyle yaptıkları tahminlerden daha iyi tahmin yapabilen modellerin
oluşturulması için öncelikle eldeki veri ve çözülmeye çalışılan problem (eBGBM değerinin tahmin
edilmesi) analiz edilmiştir. Öncelikle veri kümesinden örnek alt kümeler seçilmiştir. Üç reklam
102
alanı ve beş ülke için seçilen veri kümelerindeki gözlem sayısı (gerçekleşen eBGBM değerleri)
Tablo 3’te gösterilmektedir.
Tablo 3. Analiz için Kullanılan Veri Kümelerindeki Gözlem (eBGBM değeri) Sayıları
Reklam alanı 1
Reklam alanı 2
Reklam alanı 3
Ülke
BR
144
250
275
DE
144
261
280
FR
144
261
280
TR
144
259
283
US
139
262
280
Tablo 3’te gösterilen her bir veri kümesinin durağan olup olmadığını anlamak için
Augmented Dickey-Fuller testi yapılmıştır. Durağan bir zaman serisinde gözlemler (bu çalışma
kapsamında gerçekleşen eBGBM değerleri) zamana bağlı değildir. Bir zaman serisi trend ya da
mevsimsel etki içermiyor ise durağan olarak nitelendirilmektedir. Durağan olmayan zaman
serilerinde ise mevsimsel etkiler, trendler ve zamana bağlı başka örüntüler gözlemlenmektedir.
Tablo 4’te gösterildiği gibi az sayıda durağan veri kümesi (Tabloda “D” ile ifade edilmiştir)
bulunurken veri kümelerinin çoğu durağan değildir (Tabloda “DD” ile ifade edilmiştir).
Tablo 4. Veri Kümelerinin Durağanlık Durumu (D: Durağan; Dd: Durağan Değil)
Reklam alanı 1
Reklam alanı 2
Reklam alanı 3
BR
D
DD
DD
DE
DD
DD
DD
Ülke
FR
DD
DD
DD
TR
D
DD
DD
US
D
DD
DD
Bu analiz ve veri üzerinde alan uzmanları ile yapılan çalışmalar sonucunda ARİMA modeli
kullanılarak zaman serisi analizi yapılmasına karar verilmiştir. ARİMA modeli, zaman serisi
verilerini analiz etmek ve tahmin etmek için kullanılmaktadır. ARİMA, “AutoRegressive
Integrated Moving Average” kavramını temsil etmektedir. ARİMA modelinde üç hiper-parametre
bulunmaktadır [6]:
•
p: Model içerisinde yer alan gecikmeli gözlemlerin sayısını belirtir; gecikme sırası olarak
adlandırılır.
•
d: Ham gözlemlerin sayısını belirtir; ayırt edilme derecesi olarak adlandırılır.
•
q: Hareketli ortalamanın penceresinin boyutudur; hareketli ortalama sıralaması olarak
adlandırılır.
Bu üç hiper-parametrenin hangi kombinasyonunun en iyi tahmin değerlerini verdiğini
belirlemek için grid arama yöntemi kullanılmıştır. En iyi kombinasyonu bulmak için her parametre
103
için aşağıdaki değerler kullanılmıştır. Tablo 4’te gösterildiği gibi hem durağan hem de durağan
olmayan veri kümeleri olduğu için d parametresi hem 0 hem de 1 değeri alacak şekilde
düzenlenmiştir.
•
p = 0, 1, 2, 3, 4, 5, 6, 7
•
d = 0, 1
•
q = 0,1, 2
En iyi kombinasyon, bir yayıncı uygulamanın üç reklam alanının beş farkı ülkedeki eBGBM
değerleri kullanılarak yapılmıştır. Her bir veri kümesinin üçte ikisi öğrenme, geri kalan üçte biri
ise test için kullanılmıştır. Elde edilen hiper-parametre (p, d, q) değerleri Tablo 5’te
gösterilmektedir. Bu hiper-parametre kombinasyonları kök ortalama kare hatası (KOKH) (root
mean square error) en az olacak şekilde oluşturulmuştur. Hesaplama için kullanılan formül aşağıda
gösterilmektedir.
Reklam alanı Reklam alanı Reklam alanı
3
2
1
Tablo 5. Deney Sonucunda Elde Edilen Hiper-Parametre Değerleri
Ülke
BR
DE
FR
TR
US
BR
DE
FR
TR
US
BR
DE
FR
TR
US
p
d
0
1
0
1
1
1
1
1
1
1
0
4
1
1
0
q
1
0
0
0
0
1
0
0
0
1
1
0
1
0
1
1
0
1
0
1
0
1
1
1
0
1
2
0
1
1
Elde edilen hiper-parametre kombinasyonları ve öğrenme için ayrılan veri kümesi
kullanılarak geri kalan test veri kümesindeki gerçekleşen değerler için tahmin yapılmıştır. Tahmin
yapılırken test kümesi için ayrılan gerçekleşen eBGBM değerleri kullanılmamıştır. Sonrasında
tahmin edilen eBGBM değerleri ile gerçek eBGBM değerleri karşılaştırılarak kök ortalama kare
104
hataları hesaplanmıştır. Aynı şekilde gerçek eBGBM değerleri ile geçmişte alan uzmanları
tarafından oluşturulan tahminler de karşılaştırılarak kök ortalama kare hataları hesaplanmıştır.
Sonuçlar Tablo 6’da gösterilmektedir.
Tablo 6. Gerçekleşen Ebgbm Değerleri ile ARİMA Modeli ve Alan Uzmanının Tahmin Değerleri
Arasındaki Kök Ortalama Kare Hataları
Reklam alanı Reklam alanı Reklam alanı
3
2
1
Ülke
BR
DE
FR
TR
US
BR
DE
FR
TR
US
BR
DE
FR
TR
US
Gerçek –
ARİMA
tahmini için
KOKH
0,171
1,895
0,831
0,154
3,941
4,601
6,628
11,315
2,639
6,211
1,940
4,533
4,914
1,615
6,104
Gerçek –
Alan uzmanı
tahmini için
KOKH
0,162
2,300
0,812
0,183
3,648
2,402
6,738
10,437
2,109
5,224
1,599
3,802
3,369
1,295
5,088
Tablo 6’da koyu ve italik olarak gösterilen sonuçlar, gerçekleşen eBGBM değerleriyle daha
az bir farkı temsil ettiği için diğerine göre daha başarılıdır. Elde edilen sonuçlara göre alan
uzmanlarının kurduğu modelin yaptığı tahminler ARİMA modeli ile elde edilen tahminlerin
çoğuna göre daha başarılıdır.
Sonuçlar ve Gelecek Çalışmalar
Reklam aracısı yazılımları, gerçekleşecek eBGBM değerlerini tahmin ederek yayıncı
uygulamaların reklam gelirlerini arttırmayı hedeflemektedir. Bu tahmin işlemi için kurulabilecek
en basit modellerden birisi geçmiş belirli bir dönemin ortalamasını tahmin olarak kullanmaktır.
Diğer taraftan zaman serisi analizi gibi yüksek hacimli veri kümelerindeki gizli örüntüleri bularak
daha iyi tahminler yapabilme potansiyeline sahip modeller bulunmaktadır. ARIMA bu
modellerden birisidir. Bu çalışma kapsamında ARIMA modeli kullanılarak eBGBM değerleri
tahmin edilmiş ve bunlar alan uzmanlarının tasarladığı modelin yaptığı sonuçlarla
karşılaştırılmıştır. Yapılan deney sonucunda alan uzmanlarının tahminleri daha başarılı çıkmıştır.
105
Gelecek çalışmalar kapsamında, elde edilen sonucun geçerliliği daha fazla veri ile test
edilecektir. Bunun yanında ARİMAX modeli kullanılarak ve tahmin başarısını arttırabilecek
değişkenler bu modele eklenerek deney yapılacaktır. Ayrıca hiper-parametre optimizasyonu
sırasında kombinasyon sayısı ve parametre değerleri arttıkça performans problemleri yaşanmıştır.
Bu problemi gidermek için optimizasyonun MapReduce dağıtık programlama modeli ile yapılması
planlanmaktadır.
TEŞEKKÜR
Bu çalışma TÜBİTAK’ın desteğiyle 3171053 numaralı proje kapsamında yapılmıştır.
KAYNAKÇA
[1] E. Llach, “System For Automatically Selling and Purchasing Highly Targeted and Dynamic
Advertising Impressions Using A Mixture Of Price Metrics,” U.S. Patent Application No. 10/767,050,
2004.
[2] B. O’kelley, “Method And System For Pricing Electronic Advertisements,” U.S. Patent
Application No. 11/006,121, 2006.
[3] A. Paunikar And M. Hochberg, “Dynamic Pricing For Content Presentations,” U.S. Patent No.
8,706,547. 22, 2014.
[4] O. Chapelle, E. Manavoglu, R. Rosales, “Simple And Scalable Response Prediction For Display
Advertising,” Acm Transactions On Intelligent Systems And Technology (Tıst), Vol. 5(4), Article 61, 2015.
[5] Ö. Kırkağaçlıoğlu, G. Giray, A. Ersin, C. Çatıkkaş, S. Koçer, T. Şeremet, M. O. Ünalır, “Bir
Reklam Aracısı Yazılımının Mimari Evrimi,” 12. Ulusal Yazılım Mühendisliği Sempozyumu, İstanbul,
2018.
[6] J. Brownlee, Introduction To Time Series Forecasting With Python, Edition V1.6, 2019.
106
Marketing and Data Analytics; Increasing Importance of Marketing
Muhammet GİRGİN
Social Sciences Vocational School, Karabük University, TURKEY
Abstract: —At the moment we are entering a new era of knowledge, we are experiencing the first
stages of a digital transformation. Businesses should be restructured in a structure suitable for the
digital age when they are moving towards Industry 4.0, which is rising on cyber physical systems.
In this period when the digital enterprises are rising, each business unit of the enterprise should be
in the effort of using these innovations in accordance with this structure and using them in the most
effective way to reach their goals. With the understanding of marketing in the modern sense, one
of the most basic functions of marketing is to provide the information about the customer demands
and expectations in order to be used in the product design and planning stage. For this reason,
marketing is not only about the sales and marketing of the products or services produced, but also
for the decision of what to produce. In this period, where the effects of the digital age are becoming
more evident, the weight of digital applications in marketing is increasing and the concept of
Digital Marketing is becoming more prominent. Innovations on the transformation of Data (which
is called New oil) into information, give enterprises more competitive advantages and help to make
more accurate decisions. In parallel with the developments in technology, the change in the market
and the business environment necessitates the marketing to be prepared for these changes and to
transform itself with this change. In this study, the analytical concept is discussed with marketing.
In addition, the relationship between the analytical concepts of marketing is tried to be explained
and the opportunities awaiting marketing are put forward. At the end of the study, it has been
concluded that this digital transformation and developments all over the world increase the
importance of marketing.
Keywords: Analytics, Marketing Analytics, Marketing 4.0, Big Data Analytics, Digital Marketing
107
REFERENCES
[1] A. C. Ramos, S. C. Fermont: Insider SEO & PPC. Jain Publishing. 2006
[2] A. Gilmore, D. Gallagher, H. Scott, E-Marketing And Smes: Operational Lessons For The
Future. European Business Review, 2007, 19(3), 234-247.
[4] B. Bulunmaz, Gelişen Teknolojiyle Birlikte Değişen Pazarlama Yöntemleri ve Dijital
Pazarlama TRT Akademi Cilt 1 / Sayı 2 / Temmuz 2016 Dijital Medya Sayısı.
[5] B. Manel, S. Dupuy-Chessa, L. Gzara, N. Mandran, C. Jeannet, BI4BI: A Continuous
Evaluation System For Business Intelligence Systems Expert Systems With Applications, Journal
Homepage: www.Elsevier.Com/Locate/Eswa, Volume 76 Pages 1-166 (15 June 2017) S:97.
[6] C. Dave, M. Patron, From Web Analytics To Digital Marketing Optimization: Increasing The
Commercial Value Of Digital Analytics, Journal Of Direct, Data And Digital Marketing Practice. July
2012, Volume 14, Issue 1.
[7] C. Gurâu, F. Duquesnois, The Website As An İntegrated Marketing Tool: An Exploratory Study
Of French Wine Producers. Journal Of Small Business And Entrepreneurship, 2011 24(1), 17-28.
[8] C. L. Bovee, ve J. V. Thill., Marketing, McGraw Hill Inc., USA, 1992.
[9] C. J. Shultz II. There İs More To Marketing: An Encore To Layton's Ring Cycle And The
Compelling Case For Marketing Systems Theory, Research And Management, Australasian Marketing
Journal 24, 2016.
[10] D. Chaffey, P.R. Smith, E-Marketing Excellence. New York: Routledge. 2013.
[11] D. Chaffey M. Patron, From Web Analyti Cs To Digital Marketing Optimization: Increasing
the commercial valueof digital analytics, Journal of Direct, Data and Digital Marketing Practice (2012) 14,
30– 45.
[12] D. Q. Chen, D. S. Preston & M. Swink, How The Use Of Big Data Analytics Affects Value
Creation İn Supply Chain Management Journal Of Management Information Systems, Journal Of
Management Information Systems / 2015, Vol. 32, No. 4, Pp. 4–39.
[13] E. Uzunoğlu, İnternet Reklamcılığında Yeni Bir Yaklaşım, Edt. Sinem Yelgen Çakır
Teknolojinin Pazarlama İletişimine Etkileri, Nobel Yay. Dağ. 2010.
[14] E. Schmidt, J. Cohen, Yeni Dijital Çağ, Çev. Ümit Şensoy, Optimist Yayınevi, 2014.
[15] F. B. Peltekoğlu, Halkla İlişkiler Nedir? İstanbul: Beta Basım Yayım Dağıtım. 2016.
108
[16]
G.
Çözen,
(2011).
Dijital
Pazarlama
Nedir?
Teknikleri
Nelerdir?
Http://Www.Dijitalmarketing. Net/2011/03/31/Dijital-PazarlamaNedir-Teknikleri-Nelerdir/, Erişim: 07
Mart 2018.
[17] G. J. Dehkordi, S. Rezvani, M.S. Rahman, F. Fouladivanda, N. Nahid, S.F. Jouya, A
Conceptual Study On E-Marketing And İts Operation On Firm’s Promotion And Understanding Customer’s
Response. International Journal Of Business And Management, 7(19), 114-124. 2012.
[18] G. Viglia, Princing, Online Marketing Behavior, And Analitics, Palgrave Macmillian
Publihed, 2014.
[19] H. J. Watson, H. W. Barbara, The Current State Of Business Intelligence, 96-97 IT Systems
Perspectıves IEEE Computer Society, 2007.
[20] İ. E. Emre, Ç. S. Erol, Veri Analizinde İstatistik mi Veri Madenciliği mi? Bilişim Teknolojileri
Dergisi, Cilt: 10, Sayı: 2, Nisan 2017.
[21] İ. Ertuğrul, Deniz, Gözde. 4.0 Dünyası: Pazarlama 4.0 ve Endüstri 4.0. Bitlis Eren Üniversitesi
Sosyal Bilimler Enstitüsü Dergisi, 2018.
[22] J. Järvinen, H. Karjaluoto, The Use Ofweb Analytics For Digital Marketing Performance
Measurement, Industrial Marketing Management, 2015.
[23] K. Nakatani, T. Chuang, (2011) "A Web Analytics Tool Selection Method: An Analytical
Hierarchy
Process
Approach",
Internet
Research,
Vol.
21
Issue:
2,
Pp.171-
186,
Https://Doi.Org/10.1108/10662241111123757.
[24] K. Suryanarayanan, S. K. Mathew, Business Analytics and Business Value: A Comparative
Case Study, Information & Management, Available Online 31 January 2018.
[25] L. Fink, N. Yogev, A. Even, Business Intelligence and Organizational Learning: An Empirical
İnvestigation Of Value Creation Processes, Information & Management 54 (2017) 38–56
[26] M. C. Cant, N. V. Rooyen, The Use of Search Engines in The Marketing of Accommodation
Establishments: Adopt or Reject The Use Of Google Adwords?, International Review Of Management And
Marketing, 2017, 7(3)
[27] M. Kubinaa, M. Varmusa, Irena Kubinova, Use of Big Data For Competitive Advantage Of
Company, 4th World Conference On Business, Economics And Management, WCBEM, Procedia
Economics And Finance 26 (2015)
109
[28] M. Kırlı, H. Gümüş, Y. Altıntop, Stratejik Yönetim Muhasebesinin Tamamlayıcısı Olarak
Stratejik Pazarlama Muhasebesi, 4. Uluslararası Balkanlarda Sosyal Bilimler Kongresi, 2012
[29] N. Köse, D. Yeğin, Dijital Pazarlamadan Fijital Pazarlamaya Geçişe Örnek Olarak Artırılmış
Gerçeklik Ve Sanal Gerçeklik Uygulamalarının Pazarlama Üzerindeki Katkılarının İncelenmesi, İstanbul
Aydın Üniversitesi Dergisi - Ocak 2018 Cilt 10 Sayı 1
[30] P. Kotler, Gary Armstrong. Principles of Marketing, 8th Ed., Prentice-Hall Inc. New JerseyUSA, 1999.
[31] P. Kotler, Marketing Management, New Jersey: Pearson Education International. 2003
[32] P. Kotler, Kartajaya, Hermawan. Setiawan, Iwan. Marketing 3.0 John Wiley & Sons
Publishing, 2010. USA
[33] P. Kotler, H. Kartajaya, I. Setiawan, Pazarlama 4.0 Gelenekselden Dijitale Geçiş, Optimis
Yayın, 2018, İstanbul
[34] Ö. Torlak, R. Altunışık, (2018). Pazarlama Stratejileri Yönetsel Bir Yaklaşım, Beta Yayınları
[35] R. Glass, S. Callahan, The Big Data Driven Business, John Wiley & Sons Publishing, 2015.
USA
[36] S. Dann, S. Dann, E-Marketing Theory And Application. New York: Palgrave Macmillan.
2011
[37] S. Ganesh, “Data Mining: Should It Be Included in The Statistics Curriculum?", The 6th
International Conference On Teaching Statistics (ICOTS 6), Cape Town, Güney Afrika, 2002.
[38] T. Anna, The Reevaluation Of Communication In Customer Aproach – Towards Marketing
4.0, Internatipnal Journal of Contemporary Management, 12(4), 124-134, 2013
[39] T. Davenport, Big Data @ Work, Türk Hava Yolları Yay. Çev; Müge Çavdar. İstanbul 2014
110
Data Analytics and Importance in Health Sector
Duygu KARABULUT1, Taner ERSÖZ2
1
2
Department of Industrial Engineering,
Department of Actuary and Risk Management, Karabük University, TURKEY
Abstract: Nowadays, as a result of the fast advancement of technology, large and complex
databases are created. Since it may be hard to process the created complex data, there is a need for
software programs. The purpose of this study is to create a substructure for data mining to be used
in the health sector and to give a quick and different perspective in deciding to reach the desired
data. With this purpose, based on the data from the operating room of a private hospital, the data
mining method was used in this study.
Keywords: Health Sector, Data Analytics, Predictive Analysis, Classification
Giriş
Sağlık sektörü, sağlığa etkileri olan her türlü ürünü arz - talep etmek, tüketmek üzere çok farklı
üretim alanlarında kurulmuş bütünleşik sistemler ile bunların içerdiği kişi, kurum, kuruluş, ürün ve
benzerleri tümünü belirtmek için kullanılan bir kavramdır.
Türkiye’de sağlık hizmetleri Cumhuriyet’in kuruluşu 1923 yılından itibaren 1982 yılına kadar
devletin sunması esas alınan bir hizmet olmuştur. Türkiye’de kamu dışında hizmet sunan sağlık
kuruluşları sayısında hızlı artış ve özel sağlık sigortası için sağlık alanının gelişmesinin yaşandığı
yıllar 1900’lerdir. Türkiye’de sağlık sektörü, 1980’lerden sonra 20 yılda yaklaşık 3 kat büyümüş,
bu büyümede kamu sektörünün payı giderek belirleyici olmuştur.
Dünya ekonomisine bakıldığında günümüzde sağlık sektörü ilk sıralarda yer almaktadır. Sağlık
sektörü için 1990’lı yıllarda yapılan toplam harcama yaklaşık 2985 milyar dolardır. Bu harcama
dünya brüt milli hasılasının yaklaşık %8’ne denk düşmektedir. Bu rakamsal boyut, binlerce yıl
öncesinde, aile üyeleri, dini örgütler ya da bazen profesyonel bir şifa verici tarafından tedavi
sunulan bir kişiyle, bir hastalık arasındaki basit ve özel bir ilişkinin, geçmiş iki yüzyıl içinde nasıl
genişlediğini ve bir sağlık sistemi tarafından kapsanan kompleks bir ağa nasıl girdiğini
111
yansıtmaktadır (WHO 2000). Sağlık sistemi performansını ele almadan önce yapılması gereken,
performans kriterlerinin geliştirilmesine temel teşkil edecek şekilde sağlık sisteminin
tanımlanması ve sınırlarının çizilmesidir. Günümüzde bir sağlık sisteminin ne olduğunu, nelerden
oluştuğunu ve nerede başlayıp nerede bittiğini tam olarak söylemek zordur. Dünya Sağlık Örgütü
(DSÖ) 2000 Raporu sağlık sistemini, temel amacı sağlığı geliştirmek, yenilemek ve sürdürmek
olan tüm aktiviteleri içerecek biçimde tanımlamıştır (WHO 2000; Murray, Frenk 2000; WHO
2001; Murray, Frenk 2001; IHSD 2000a; WHO 2000a).
Günümüzde sağlık s stemler nde b rçok g rd , süreç ve çıktılar söz konusudur. B nlerce
hastalık, teşh s ve tedav yöntem , laç, tıbb malzeme ve şlemler mevcut olmakla b rl kte, sağlık
s stem b rçok kurum, kuruluş, k ş , kaynak ve f nansman model n çerd ğ nden dolayı bunlar
arasındak
planlama, koord nasyon ve yönet m faal yetler
profesyonel yönet m anlayışı
gerekt rmekted r.
Doğru karar vermede, ver ler n toplanması, hazırlanması, anal z ed lmes
ve doğru
yorumlanması çok öneml d r. Ver len kararın doğruluğu b lg b r k m ne bağlı olduğu kadar ver n n
kaynağının yeterl l ğ le de l şk l d r.
Tıp alanındak d j tal ver ler n artışı b rçok sorunları beraber nde get rm şt r. Ver madenc l ğ
metadoloj s günümüzde b rçok farklı sektörle b rl kte tıp alanında da oluşan b rtakım açığa çıkan
problemlerden dolayı g derek yaygınlaştığı görülmekted r.
Cerrah müdahaleler büyük m ktarda r sk oluşturab leceğ
ht mal ne karşılık d kkatle
planlanmalıdır. Cerrah g r ş m n uygun seç lmes ç n hasta değerlend rmes temel alınmalıdır.
Müdahale seç m nde hastanın öyküsü, f z ksel durumu ve tanısal ver ler n önem kadar
müdahalen n hasta ç n r skler ve yararları da son derece öneml d r. Müdahale seç m nde, başvuru
bulguları, tanısal test ve d ğer kaynaklardan elde ed len b lg ler de göz önüne alınmalıdır.
Günümüzde cerrah müdahaleler n en öneml ler arasında yer alan Ortoped ve Travmatoloj ,
nsanın doğumundan ölümüne kadar olan kas ve skelet s stem hastalıklarının neredeyse
tamamıyla dolaylı ya da d rek olarak lg lenen b l m dalıdır. Ortoped , kel me anlamı düzgün
çocuktur. Travmatoloj se travma sonucu yaralanmalara bakan bölüm olarak adlandırılır.
Bu çalışmada özel b r hastanen n 2010- 2014 yılları arasında gerçekleşt r len amel yatların
dağılımı Şek l 1’de ver lm şt r. Genel ortoped n n yanında artroskop k cerrah , travma cerrah s ve
eklem protezler amel yatların da stat st k dağılımı ver lmekted r.
112
2%
1% 3%
1
12%
39%
2
3
4
27%
5
16%
6
7
Şek l 1. 2010- 2014 Yılları Arası Amel yatların Dağılımı
1.Genel ortoped ; 2.D z ve kalça protez ; 3.Artroskop k cerrah ; 4.Travma cerrah s ; 5.Omurga
cerrah s ; 6.Tümör cerrah s ;7.Tümör cerrah s
Ortoped ve Travmatoloj , teknoloj ve b l msel lerlemeler n ışığında en hızlı gel şen, değ şen
ve hatta kapsamı artan dalların başında gelmekted r.
Alt Dalları:
• Artroplast cerrah s (eklem protezler )
• Spor travmatoloj s
• Boy uzatma ve bacak eş ts zl kler
• Çocuk ortoped ve travmatoloj s
• El cerrah s ve m krocerrah
• D z cerrah s ve artroskop k cerrah
• Omuz ve d rsek cerrah s
• Ortoped k onkoloj
• Ayak ve ayak b leğ cerrah s
• Kem k İlt hapları (Osteomyel t) tedav s
113
Bu çalışmada özel b r hastane amel yathane ver ler kullanılarak ver madenc l ğ uygulamaları
gerçekleşt r lm şt r. Hastanen n gelecek aylardak hasta yoğunluklarının, amel yathane çeş tler n n
c ns yet bazlı dağılımı esas alınarak tahm n ed lmes nde IBM SPSS Modeler üzer nde çalışılmış
ve böylece en kest r mc stat st k ver ler n elde ed lmes esas alınmıştır.
Literatür
Selma Altınd ş ve İlknur Kıran Morkoç’un yapmış oldukları 2018 yılındak çalışmada Sağlık
H zmetler ndek Büyük Ver lere değ n lmekted r. Çalışmanın amacı, sağlık h zmetler nde büyük
ver ve kullanım alanları hakkında b lg vermekt r. Sonuç olarak Sağlıkla lg l ver ler n devasa
m ktarlara ulaşması, bu ver ler n geleneksel ver
zorlaştırmış ve büyük ver
şleme yöntemler tarafından şlenmes n
kavramının sağlık h zmetler ne g rmes ne neden olduğuna
varılmaktadır. Daha önce geleneksel ver şleme yöntemler le depolanamayan, yönet lemeyen ve
anal z ed lemeyen yüksek hac ml , hızlı ve çeş tl ver kümeler n n anlamlı ve değer yaratacak
sonuçlara dönüşmes Büyük Ver le mümkün olab leceğ görüşü savunulmaktadır. Ülkem zde de
sağlık s stem n n performansını artırmak amacıyla büyük hac mlerdek sağlık ver setler n
toplamak ve anal z etmek üzere Büyük Ver Araştırma Enst tüler n n kurulması öner lmekted r.
Sezg n Irmak, Can Den z Köksal, Özcan As lkan’ın yapmış oldukları 2012 yılındak çalışmada
hal hazırda şleyen b r hastane ver tabanında bazı öneml ver madenc l ğ tekn kler le hasta
yoğunluklarının tahm n ed lmes uygulamaları yapılmış ve sonuçları karşılaştırmalı olarak
aktarılmıştır. Bu çalışmada b r hastane ver tabanı kullanılarak ver madenc l ğ uygulamaları
gerçekleşt r lmekted r. Hastanen n gelecek aylardak hasta yoğunluklarının tahm n ed lmes nde
üstel düzgünleşt rme, ARIMA ve yapay s n r ağları tekn kler önce kend ç nde farklı modellerle
karşılaştırılmış sonrasında da her tekn ğe a t en y modeller kend aralarında kıyaslanmıştır.
Böylece en kest r mc model bel rlenmeye çalışılmıştır. Çalışmada kest r mc anal z olarak
gelecektek hasta yoğunluklarının tahm n ed lmes amaçlanmış ve üç farklı ver madenc l ğ
tekn ğ ve bunların da kend ç nde farklı modeller üret lerek gelecektek hasta yoğunluklarının
tahm n ed lmes ve bu konuda en y modeller n bel rlenmes amaçlanmıştır. Üstel düzgünleşt rme
yöntemler arasında en kest r mc model W nters Add t ve model olmuştur. ARIMA süreçler
ç nde en kest r mc model ARIMA(3,1,0)(1,0,0)12 model olmuştur. Yapay s n r ağları yöntemler
arasında se en kest r mc model Prune yöntem yle elde ed len model olmuştur. Yapay s n r ağları
modeller arasında ver ye daha fazla uyum gösteren modeller olmasına rağmen bunlarda
114
aşırıöğrenme problem n n gerçekleşt ğ görülmüştür. Her yöntem n en kest r mc modeller n n
b rb rler yle kıyaslanması, uyum y l ğ kr terler ve modeller n tahm n değerler
le hastane
ver tabanından elde ed len gerçekleşen hasta sayısı değerler n n karşılaştırılması yöntemler
kullanılarak gerçekleşt r lm şt r. Her k konuda da W nters Add t ve üstel düzgünleşt rme model
en kest r mc model olmuştur. Uyum y l ğ kr terler bakımından ve lk 7 aydak tahm nler n
gerçeğe yakınlığı bakımından ARIMA(3,1,0)(1,0,0)12 model
k nc en y olsa da 8. ve 9.
aylardak tahm n değerler kötüye g tm ş ve yapay s n r ağları model bu aylarda W nters Add t ve
üstel düzgünleşt rme model nden sonra k nc en y tahm nler gerçekleşt rm şt r. Gerçekleşen
sayıların tahm nlere oldukça yakın olması bu tekn kler n hastanen n yoğunluk tahm nler ç n
kullanılab leceğ n göstermekted r. Ayrıca ver madenc l ğ tekn kler kullanılarak sağlık sektörü
ver tabanları veya ver ambarlarından b rçok amaç ç n faydalı b lg ler n elde ed lmes de mümkün
gözükmekted r.
Al Serhan Koyuncug l, Nerm n Özgülbaş’ın 2009 yılında yapmış oldukları çalışmada sağlıkta
Ver Madenc l ğ n n kullanımı konusunda b r altyapı oluşturmak ve sağlık profesyoneller ne sağlık
sektöründe Bu çalışmada sırasıyla Ver Tabanlarında B lg Keşf , Ver Ambarı, Ver Madenc l ğ ,
İş Zekası ve Ver Madenc l ğ Yöntemler konularında tanımlayıcı b lg lere yer ver lmekte;
ülkem zdek sağlık sektöründe öncel kl konu ve sorun alanları d kkate alınarak Ver Madenc l ğ
uygulamalarına örnekler ver lmekted r. Geleceğ n sayısal karar verme ve ş zekası yöntem olan
Ver Madenc l ğ n n konunun uzmanı k ş ler tarafından sağlık sektöründe kullanımı, sağlık
h zmetler n n daha etk n sunumu, kaynakların daha ver ml
kullanımı ve b l msel,
karşılaştırılab l r, şeffaf b lg er ş m açısından öner lmekted r.
Şebnem Aslan, Mete Sezg n, Selçuk Burak Haşıoğlu’nun 2008 yılında yapmış oldukları
çalışmada özel sağlık kuruluşlarında müşter memnun yet ne etk eden faktörler n ve müşter ler n
sağlık kuruluşu terc hler ndek ölçütler araştırılmaktadır. Araştırma, Konya l nde yed özel sağlık
kuruluşundan yararlanan 200 katılımcıyla gerçekleşt r lm şt r. Araştırmada Anal t k H yerarş
Sürec (AHS) yöntem nden yararlanılmıştır. Araştırmanın sonucunda, müşter memnun yet ne etk
eden faktörlerden en etk l ölçütün, algılanan kal te olduğu tesp t ed lm şt r. D ğer ölçütler
sırasıyla; f yat, kolaylık, referans ve yakınlık bulunmuştur. Ayrıca çalışmada, doktor h zmet n n,
müşter memnun yet n en yüksek düzeyde etk leyen değ şken olduğu tesp t ed lm şt r.
Mehtap Çakmak, M. Kemal Öktem, Uğur Ömürgönülşen’ n yapmış oldukları 2008 yılındak
çalışmanın amacı, genelde Türk kamu hastaneler n n etk nl k sorununu rdelemek ve özelde se
115
T.C. Sağlık Bakanlığı’na bağlı kadın doğum hastaneler n n tekn k etk nl kler n ölçmekt r. İk nc
basamak sağlık h zmet sunan ve b rden çok g rd s ve çıktısı bulunan kadın doğum hastaneler n n
tekn k etk nl kler n n ölçümünde, g rd ve çıktı çeş tl l ğ sorunu karşısında b rden fazla g rd ve
çıktıyı aynı anda hesaba katarak ölçüm yapan Ver Zarflama Anal z (VZA) tekn ğ kullanılmıştır.
Yapılan ölçüm sonucunda, araştırma kapsamındak hastaneler n yaklaşık 1/3’nün etk n, 2/3’ünün
se etk ns z faal yet gösterd ğ saptanmıştır.
G zem Gülsev n, Ayça Hat ce Türkan’ın 2012 yılında yapmış oldukları çalışma,
Afyonkarah sar’dak Sağlık Bakanlığı’na bağlı hastaneler n etk nl k düzeyler n n ver zarflama
anal z (VZA) yöntem le bel rlenmes n amaçlamaktadır. VZA, parametr k olmayan b r etk nl k
ölçüm yöntem olup d ğer etk nl k ölçüm yöntemler ne göre daha gerçekç ve doğru sonuçlar
ortaya çıkarır. Bu nedenle çalışmada anal z ç n VZA tekn ğ terc h ed lm şt r. Çalışma,
Afyonkarah sar İl Sağlık Müdürlüğünden sağlanan Afyonkarah sar’dak Sağlık Bakanlığına bağlı
hastaneler n 2011 yılına a t ver ler n çermekted r. Hastane yönet m n n g rd ler üzer nde kontrol
gücü vardır, ancak çıktılar üzer nde kontrol gücü oldukça zordur. Bu nedenle çalışmada g rd ler
m n m ze etmey amaçlayan, ölçeğe göre sab t get r varsayımına dayanan g rd yönlü Charnes
Cooper Rhodes (CCR) model kullanılmıştır. Kurulan model W nQSB ve EMS programlarından
yararlanılarak çözülmüştür. Yapılan anal z sonucunda 8 hastane %100 etk nl k skoru le etk n
olarak bulunmuştur.
Şeyda Gür, Buse Uslu, Tamer Eren, Nesr n Akca, Al Yılmaz, Seda Sönmez’ n yapmış oldukları
2018 yılındak çalışmada, amel yathaneler n performanslarının artırılmasında etk l olan kr terler n
bel rlenmes
ç n çok ölçütlü karar verme yöntemler nden Anal t k Ağ Sürec
yöntem
kullanılmıştır. Kr terler arasındak bağımlılıkları, etk leş mler ve ger b ld r mler d kkate alan bu
yöntem le kr terler n b rb r üzer ndek l şk ler ncelenm şt r. Çalışmada kullanılan Anal t k Ağ
Sürec yöntem
le amel yathane performanslarının artırılmasında etk l olan kr terler n önem
dereceler hesaplanmıştır. Bu önem dereceler ne göre hang kr ter n performans üzer nde ne derece
etk s olduğu göster lm şt r. Elde ed len sonuçlara bakıldığında; ş gören kr ter , mal yet kr ter ve
öğrenme ve büyüme kr ter n n ön plana çıktığı görülmekted r.
Müberra Terz ’n n 2018 yılında yapmış olduğu çalışmada ver madenc l ğ hakkında b lg
ed n p ver madenc l ğ metotlarının ülkem z sağlık sektöründe nasıl kullanıldığına ve hang
alanlarda kullanılab leceğ ne değ n lm şt r. Bu derlemede ver madenc l ğ hakkındak b lg ler tek
çatı altında toplanmaya çalışılmış, ülkem z sağlık sektöründe ver madenc l ğ uygulamalarına ve
116
kullanım alanlarına göz atılmıştır. Derlemedek
örnek uygulamalar ülkem zde yapılan
çalışmalardan seç lm ş olup hastalık r sk üzer ne yapılan çalışmaların daha fazla olduğu
gözlemlenm şt r. Hastalık r sk üzer ne olan çalışmalardan sonra laç kullanımlarıyla lg l olan
çalışmalar d kkat çekmekted r. Ver madenc l ğ n n sağlık sektöründek çalışmaları hastalık r sk
ve laç dozu le sınırlı olmayıp sağlık sektöründe çalışanların beklent ler derleme çer s nde
ver lm şt r. Çalışmaların sonuçları yüksek başarı oranı çermekte olup ver madenc l ğ n n sağlık
çalışanlarına ver ler yorumlamalarında yardımcı b r araç olab leceğ düşünülmekted r.
Selma Altınd ş’ n 2018 yılında yapmış olduğu çalışma le hasta memnun yet , sürekl
gel şt rme, etk l l k, ver ml l k ve hasta güvenl ğ kavramları üzer nden günümüzde öneml b r
konu hal ne gelen büyük ver
le sağlık h zmetler kal tes
l şk s n n değerlend r lmes
amaçlanmaktadır. Sonuç olarak, büyük ver ler n kullanımında deney ms zl k, anal t k gel şt rme
mal yet g b b rtakım zorlukların bulunmasına rağmen büyük ver teknoloj ler n n ben msenmes ,
uygulanması ve kullanılmasının, sağlık h zmetler nde olumlu b r etk ye sah p olacağı
düşünülmekted r. Bu anlamda sağlık h zmet sunan kurumlara, büyük ver anal z ç n gerekl
kaynaklara (bel rl teknoloj , anal t k yöntemler vs) ve ver bütünleşt rmes ne yönel k yatırımlar
yapmaları öner lmekted r.
Hüdaverd B rcan, Sel m Çam’ın 2016 yılında yapmış oldukları çalışmada, Cumhur yet
Ün vers tes Hastanes ’ne 2011 yılında başvurmuş olan hastaların 2006-2011 arasındak kayıtlar,
hasta başvuru davranışlarının bel rlenmes amacıyla ncelenm şt r. Oluşturulan ver set yasalar
tarafından yet şk n sayılan 18 yaş le emekl l k sınırı olan 65 yaş arasında bulunan hastalara
nd rgenm şt r. Böylece ver set 78.239 hastanın hastane ver tabanından alınan ver ler
le
oluşturulmuştur. Çalışmanın amacı, hastaların ver ler n n bulunduğu çok boyutlu b r ver tabanının
kümeleme anal z yöntemler yle ncelenmes ve ver madenc l ğ yöntemler le çok boyutlu ve
büyük hac mlerdek ver tabanlarında başarılı sonuçlar üret lm şt r.
Sabr Erdem, Güz n Özdağoğlu’nun yapmış olduğu 2008 yılındak çalışmalarında bel rl b r
dönem boyunca Ege Bölges ’ndek b r araştırma ve uygulama hastanes n n ac l serv s ne
başvuruda bulunan 214 b n hasta ver s ele alınarak, Ver Madenc l ğ nde sıklıkla kullanılan
b rl ktel k kuralı yöntem yle, ver set ndek g zl ancak anlamlılık çeren l şk ler ortaya
çıkarılmaya çalışılmıştır. Çalışma sonuçları, bölgesel özell kler taşıyab leceğ düşünülen ac l
serv slere hastaların başvuru nedenler ve hasta prof ller açısından b r f k r vermekte ve ac l serv s
117
bölümler n n yen den yapılanma çalışmalarına da farklı b r açıdan yol göstererek katkıda
bulunmaktadır.
Uygulama
Bu çalışmada özel bir hastane ameliyathane verileri alınarak veri madenciliği uygulamaları
gerçekleştirilmektedir. Hastanenin gelecek aylardaki hasta yoğunluklarının, ameliyathane
çeşitlerinin cinsiyet bazlı dağılımı ve ameliyatı gerçekleştiren doktorun ünvanı baz alınarak
değişkenlerin önem sırası ve hangi istatistiklerin etkilediği belirlenmektedir.
Bunun sonucunda ortopedi bölümündeki birden fazla rahatsızlığın cinsiyete ya da diğer
değişkenlerden hangilerine bağlı olduğu istatistiğine varılmaktadır. Tanımlayıcı istatistikler
aşağıdaki gibi doktorun ünvanı, ameliyatın türü, hastanın cinsiyeti olarak tanımlanmaktadır.
Ver Madenc l ğ model önces nde ver n n modele hazırlanması gerek r. Aşağıdak tabloda
ver n n kal tes n n y olduğu ve modele hazır olduğunu göster lmekted r.
Sınıflayıcı Quest algoritma ekran görüntüsü Şekil 1’de verilmiştir.
118
Şek l 1.Sınıflayıcı Model Quest Karar Ağacı Algor tması
Yukarıda görüldüğü gibi sonuç değişkeni, ameliyatı etkileyen en önemli değişken doktorun
ünvanı olduğu görülmektedir. Diğer değişkenler Quest karar ağacı algoritmasında sonucu
etkilememektedir.
Operasyonlarda yüksek oranda profesör doktor tercih edildiği ve vakaların %24 oranında
kemik kırığı operasyonu olduğu istatistiğine ulaşılmaktadır.
Sınıflayıcı Chaid algoritma ekran görüntüsü Şekil 2’de verilmiştir.
119
Şek l 2.Cha d Karar Ağacı Algor tması
Sonuç değ şken amel yatı etk leyen en öneml değ şken burada da doktor ünvanı olduğu
zlenmekted r. İk nc sırada c ns yet faktörünün etk s ver lmekted r.
120
İl şk anal z yukarıda graf ksel ver lmekted r. Kalın ç zg ler yüksek kolerasyonu, nce
ç zg ler düşük kolerasyon göstermekted r. Graf kte de görüldüğü g b
kem k kırığı
operasyonlarının yüksek kolerasyon gösterd ğ sonucuna varılmaktadır.
Yukarıdak graf kte doktor ünvanı değ şkenler alınarak, her b r operasyon ç n ayrı ayrı
stat st k ver lmekted r. Genel ortalamaya bakıldığında operasyonda terc hler n profesör doktor
yönünde olduğu görülmekted r.
Sonuç ve Öner ler
Bu çalışmada bir özel hastanenin ortopedi ameliyatları veri madenciliği ile araştırılmıştır.
Araştırma sonucuna göre, erkeklerde ortoped operasyonların kadınlar le karşılaştırıldığında
121
sonucunda yüzdel k olarak daha fazla yapıldığı görülmekted r. Aynı zamanda doktor ünvanının da
operasyon çeş d nde ve sayısında büyük farklılık yarattığı sonucuna varılmaktadır.
Elde ed len ver ler, hasta prof l n n gelecek aylar ç n oluşturulmasında öneml rol alacaktır. Bu
bağlamda amel yat planlaması yapılması ve ster l ürün hazırlama süreçler nde de per yotların
bel rlenmes nde büyük katlı sağlayacağı kaçınılmazdır. C ns yet bazlı elde ed len ver ler n
dağılımındak kök nedene n ld ğ nde çok daha farklı stat st k bulgulara rastlanab leceğ
öngörülmekted r.
Amel yathane süreçler ndek b rçok bel rs zl ğe neden olan büyük ve anlaşılması güç ver ler,
şlenmes ve kullanılab l r hale get r lmes sonucunda anlam kazanarak b rçok stat st k bulgu elde
etmem ze olanak sağlamaktadır. Bu bulgular zaman çer s nde süreçler n doğru ve daha hızlı
lerlemes ne yardımcı olacağı öngörülmekted r. Hasta prof l n n bel rlenmes , memnun yet
dereces n
arttırmakta da oldukça öneml d r. Sağlık sektöründek
operasyon başarılarını
arttırmanın temel nde hasta prof l n n rolü bu açıdan kes şmekted r. Bu nedenle stat st k bulgulara
operasyonun her alanında daha fazla önem ver lmes gerekt ğ düşünülmekted r.
Hastaların ver ler n n bulunduğu çok boyutlu b r ver tabanının ncelenmes ve ver madenc l ğ
yöntemler
le çok boyutlu ve büyük hac mlerdek ver tabanlarında başarılı sonuçlara
ulaşılmaktadır.
KAYNAKÇA
[1] Aydın, S nan; Özkul, Ekrem (2015), Data M n ng and An Apl cat on n Anadolu Un vers ty Open
Educat on System, Journal of Research n Educat on and Teach ng, 4.
[2] Baykal, Abdullah (2006), Appl cat on F elds of Data M n ng, D.Ü. Z ya Gökalp Eğ t m Fakültes
Derg s , 7, 95-107.
[3] Irmak, Sezg n; Köksal, Den z Can; As lkan, Özcan (2012), Pred ct ng Future Pat ent Volumes of
The Hosp tals By Us ng Data M n ng Methods, Internat onal Journal of Alanya Faculty of Bus ness, 4, 101114.
[4] Doğan, Onur (2017), Türk ye’de Ver Madenc l ğ Konusunda Yapılan L sansüstü Tezler Üzer ne
B r Araştırma, Gaz Ün vers tes İkt sad ve İdar B l mler Fakültes Derg s , 19/3, 929-951.
[5] Seker, Sad Evren (2015), Data M n ng on Soc al Networks, http://www.YBSAns kloped .com , 2,
1-5.
122
[6] Koyuncug l, Al Serhan; Özgülbaş, Nerm n (2009), Data M n ng: Us ng and Appl cat ons n
Med c ne and Healthcare, Data M n ng: Us ng and Appl cat ons n Med c ne and Healthcare, B l ş m
Teknoloj ler Derg s , 2.
[7] J. J. Fr ed (2006), “Data M n ng For Qual ty Care”, Commun Oncol, 3(1), 51.
[8] I. E. Allen (2006), C. A. Seaman, “Data M n ng For Qual ty”, Qual ty Progress, 39 (2), 70-74.
[9] A. M lley (2000), “Healthcare And Data M n ng”, Health Management Technology. 21(8), 44-46.
[10] Ma, Y., et.al. (2001). Target ng the R ght Students Us ng Data M n ng. The Seventh ACM
SIGKDD Internat onal Conference on Knowledge D scovery and Data M n ng. (pp.457–464), USA: San
Franc sco, Cal forn a, 26-29 Ağustos.
[11] Irmak, S. (2009), Ver Madenc l ğ Yöntemler le Sağlık Sektörü Ver tabanlarında B lg Keşf :
Tanımlayıcı ve Kest r mc Model Uygulamaları, Akden z Ün vers tes S.B.E. Yayımlanmamış Doktora
Tez .
[12] Kadılar, C. (2005), SPSS Uygulamalı Zaman Ser ler Anal z ne G r ş, B z m Büro Basımev ,
Ankara.
123
Program Development for Cost Calculation in Different Hole Drilling Operations
Tugay ÜSTÜN, Yakup TURGUT
Department of Manufacturing Engineering, Gazi University, TURKEY
Abstract: Drilling operation is one of the most frequently used process in manufacturing industry.
Especially in machine manufacturing sector, drilling operations take one third of the total
production time. Drilling operation cost a lot, not only because of taking some serious time of total
production time, but also because of the drill usage for this process. In this study, the cost
comparison of three different methods evaluated for drilling operation in CNC machines. These
methods are drilling cycle (G81), high-speed peck drilling cycle (G73) and deep hole peck drilling
cycle (G83). The experiments were performed according to the cutting parameters suggested by
the cutting tool company and the machining times measured in these three different methods. A
novel program coded on Microsoft Visual Studio 2017 C#, which is able to calculate from machine
amortization to workmanship, the whole process cost. Process costs can be calculated according
to the number of holes in these different methods, through this program. Furthermore, drilling
operation costs can be calculated for different cutting parameters too.
Keywords: Drilling, Cost Accounting, Programming
REFERENCES
[1] S. Kalpakjian And S. R. Schmid, “Manufacturing Engineering And Technology,” Prentice Hall,
Vol. 8th Editio, 2009.
[2] İ. Çiftçi And H. Gökçe, “Ti6al4v Titanyum Alaşımının Delinmesinde Delme Yönteminin
Aşınmaya Etkisinin İncelenmesi,” J. Polytech., Vol. 0900, Pp. 0–2, 2018.
[3] H. L. Tonshoff, W. Spintig, A. Neises, And W. Konig, “Machining Of Holes Developments in
Drilling Techonolgy,” Ann. Cırp, P. 43, 1994.
[4] S. Yagmur, A. Acir, U. Seker, and M. Gunay, “An Experimental Investigation of Effect of
Cutting Parameters On Cutting Zone Temperature in Drilling,” J. Fac. Eng. Archit. Gazi Univ., Vol. 28,
No. 1, Pp. 1–6, 2013.
124
[5] N. Bıçakcı, “Ti6al4v Titanyum Alaşımının Delinebilirliğinin Araştırılması,” Karabük
Üniversitesi, 2015.
[6] S. Sharif And E. A. Rahim, “Performance of Coated and Uncoated Carbide Tools When Drilling
Titanium Alloy Ti6al4v,” J. Mater. Process. Technol., Vol. 185, Pp. 72–76, 2007.
[7] P. K. Shetty, R. Shetty, D. Shetty, F. Rehaman, And T. K. Jose, “Machinability Study on Dry
Drilling Of Titanium Alloy Ti6al4v Using L9 Orthogonal Array”,” Procedia Mater. Sci., Vol. 5, Pp. 2605–
2614, 2014.
[8] F. Bıçakçı And A. Tekin, “Sanayi İşletmelerinde Faaliyet Tabanlı Maliyet Sistemine Geçiş
Çabaları ve Bir Uygulama,” Selçuk Üniversitesi, 2006.
[9] S. Baysan And M. B. Durmuşoğlu, “Değişen Rekabet Koşullarında Değişmeyen Maliyet
Muhasebesine Yeni Bir Soluk: Yalın Maliyet Muhasebesi,” Altı Sigma Yalın Konf., Pp. 1–5, 2008.
[10] U. Pamukoğlu And C. Göloğlu, “Delik İşlemleri için Maliyet Merkezli Bir Sistem,” J. Eng.
Sci., Vol. 10, Pp. 31–39, 2004.
[11]
“Makine
Takım
Endüstrisi
A.Ş.”
[Online].
Available:
Https://Www.Makinatakim.Com.Tr/Hss-Din-338-Silindirik-Sapli-Kisamatkap-Ucu-Haddeli/.
[12] G. Meral, “Aısı 1050 Malzemenin Delinmesinde Delme Parametrelerinin Kesme Kuvvetleri
ve Delik Kalitesi Üzerindeki Etkisinin Araştırılması,” Gazi Üniversitesi, 2010.
[13] O. E. Çelik, “Farklı Yazılım Sistemleri Kullanarak Bir Yolcu Uçağı Panelinin Üretimi için
Otomatik Perçin Tezgâhı NC Kodlarının Oluşturulması ve Karşılaştırılması,” Hacettepe Üniversitesi, 2014.
[14] P. Smid, Cnc Programming Handbook, 3rd Ed. New York: Industrial Press Inc., 2008.
[15] M. Akkurt, Talaş Kaldırma Yöntemleri ve Takım Tezgâhları. İstanbul: Birsen Yayınevi, 2004.
125
Multi-Objective Optimization of Hard Turning: Non-Dominated Sorting Genetic
Algorithm-II Approach
Ahmet KOCATÜRK, Bülent ALTUNKAYNAK
Department of Statistics, Gazi University, TURKEY
Abstract: Multi-objective optimization problems allow multiple purpose to be simultaneously
optimized. The nondominated sorting genetic algorithm II (NSGA-II), which is one of the most
effective multi-objective heuristic methods in the solution of multi-objective optimization
problems, is widely used in the literature. NSGA-II obtains a Pareto optimal solutions, known as
a set of dominant solutions without requiring any prior knowledge in one run. The NSGA-II is
more useful than the classical genetic algorithm, minimizing the computational complexity by
calculating the fast dominated sorting approach and the crowded distance without having to repeat
for each solution. In this study, NSGA-II method was used to optimize the cutting parameters of
hard materials turning. In the experimental studies, the regression models based on the cutting
velocity, feed rate and depth of cut parameters represent three different objective functions. This
optimization problem, which has five objective functions with three variables, has been discussed
by NSGA-II method. The optimal solution of these functions is to use the NSGA-II method to find
the most suitable set of Pareto solutions. The solutions obtained by using NSGAII method have
been found to be successful in multi-objective optimization problems. In addition, decision makers
from the optimal solutions can choose the most suitable solution according to their importance in
the objective functions.
Keywords: Multi-Objective Optimization, NSGA-II, Pareto Solution Set, Hard Turning
Introduction
The turning of hard materials in the machinery industry changes the operational cost due to various
parameters. It is also desirable to optimize surface roughness in turning. Thus, the parameters used
in the turning of hard materials could be linked to a common functionality and could be optimized
126
based on the economic criteria. This kind of optimization problems with multiple purposes is very
common in real life. Simultaneous optimization of multiple purposes is called multi-objective
optimization problems [1].
The solution of multi-objective problems is more complex and difficult than single-objective
problems. The local minimum or local maximum points are mathematically the best solutions for
deciding on optimization problems with a single purpose. However, it is not possible to achieve
the best solution value in optimizing conflicting goals [2].
Therefore, intuitive methods such as artificial neural networks, genetic algorithms or simulated
annealing are widely used in the solution of multi-objective problems [3-5]. However, these
methods need to be repeated many times to find each solution. As a result, the computational
complexity of classical heuristic methods is considerably greater for multi-objective optimizations.
Multi-objective heuristic methods have been developed to prevent this [6-10].
In literature studies, one of the most effective methods for solving multi-objective optimization
problems is the Non-Dominated Sorting Genetic Algorithm-II (NSGA-II) [11-15].
NSGA-II finds a set of non-dominated solutions (Pareto solution set) in single run without
requiring any prior knowledge. The basic operating principle of the algorithm is based on the fast
dominant sequence and to calculate the crowded distance according to the order of dominance in
determining the Perato surface. First, a randomly selected starting population is generated. Then,
genetic algorithm operators are determined and a new generation population is obtained. The
current and new population are combined and sorted by dominant sequence and crowded distance.
Individuals who determine the pareto surface are selected and the Pareto solution set is obtained.
The search process of the algorithm continues until the generation number.
In this study, NSGA-II method was used to optimize the cutting parameters of hard materials
turning. After some preliminary studies at [16] used in the study of hardened steel turning,
parameters were estimated for two different hardness levels. In experimental studies, regression
models based on cutting velocity, feed rate and depth of cut parameters represent three different
objective functions. These functions are: tangential force (P1), axial force (P2) and radial force
127
(P3). In addition, the surface roughness (R) and tool life (T) models are also to be optimized
simultaneously. This optimization problem, which has five objective functions with three
variables, has been discussed by NSGA-II method.
The optimal solution of these functions is to use the NSGA-II method to find the most suitable set
of Pareto solutions. Pareto optimal solution set was obtained for problems. According to the results
obtained from the study conducted by [1], it was shown that the results obtained using NSGA-II
method were better than the results obtained by genetic algorithm.
Methodology
Non-Dominated Sorting Genetic Algorithm-II (NSGA-II) is an effective heuristic method that
finds Pareto optimal solutions for multi-objective optimization [16]. The NSGA-II algorithm was
developed by revising the deficiencies of the NSGA algorithm developed by [9]. The NSGA
examines whether individuals are dominating each other for all purposes. For this comparison
𝑂(𝑀𝑁 ) calculation complexity is required. Where 𝑀 is the target number and 𝑁 is the number
of population. In the NSGA-II algorithm, computational complexity was reduced to 𝑂(𝑀𝑁 ).
When looking at the dominance of individuals for this process, instead of comparing with all
populations, the first surface in the Pareto solution compares with the individuals in the solution.
Since the NSGA-II algorithm has low computational complexity and takes into account
effectiveness, it is implemented in many areas.
NSGA-II is designed based on genetic algorithm. In addition to the steps of the genetic algorithm,
the dominant sequence and crowded distance calculation are also applied. First, the initial
population is created in NSGA-II. Solutions from the initial population are ranked according to
their superiority in the Pareto solution. In this ranking, the fast dominant sorting method is used.
Fast Non-Dominated Sorting Approach
In order to identify individuals in a population at the dominant surface in the first row, the
individual dominance of each solution is compared with the other solutions in the population. This
128
process continues for all solutions on the first dominant surface. For the second-order dominant
surface solution, the first-order solutions are ignored and the same process is repeated. Thus,
individuals are classified into different sets of dominance according to their degree of dominance.
The number of dominant and the dominant set of solutions is determined for each 𝑝 element in the
𝑃 population. The dominance value of each element on the first dominant surface is set to zero. It
is then compared with other elements in the solution set. After that, the element with the dominance
number zero is placed in another set of 𝑄. The 𝑄 set forms the second dominant surface. A
comparison is made with the elements in the 𝑄 set and the process continues in this way. The fast
non-dominated sorting algorithm is as follows:
Step 1: For each 𝑝 ∈ 𝑃;
𝑆 = ∅, The solution set in which the 𝑝 solution is dominant is defined.
𝑛 = 0, The number of solutions that dominate the 𝑝 solution is defined.
For each 𝑞 ∈ 𝑃;
If 𝑝 solution is dominant in 𝑞 solution, 𝑞 solution is added to a set of solutions where 𝑝 solution is
dominant, 𝑆 = 𝑆 ∪ {𝑞}.
If 𝑞 solution is dominant in 𝑝 solution, the dominance counter for the solution 𝑝 is increased by
one, 𝑛 = 𝑛 + 1.
If 𝑛 = 0, there is no solution to the 𝑝 solution and 𝑝 solution belongs to the first surface, 𝑝
=
1. The solution 𝑝 is added to the surface 1 and the first surface is updated, 𝐹 = 𝐹 ∪ {𝑝}.
Step 2: For each 𝑝 ∈ 𝑃; repeat Step 1.
Step 3: The surface counter is taken as 1, 𝑖 = 1.
Step 4: As long as the 𝑖. surface is different from the null set, 𝐹 ≠ ∅, the following process is
repeated:
For the surface (𝑖 + 1), the cluster where the elements are collected is taken as a null, 𝑄 = ∅.
For each 𝑝 ∈ 𝐹 and 𝑞 ∈ 𝑆 ;
Reduction of the dominance counter, 𝑛 = 𝑛 − 1.
If 𝑛 = 0, no solution on the sequential surface dominant the q solution. 𝑞
updated, 𝑄 = 𝑄 ∪ {𝑞}.
Surface counter is incremented by 1, 𝑖 = 𝑖 + 1.
129
= 1 and 𝑄 set is
New surface is assigned to 𝑄 set, 𝐹 = 𝑄.
Because the dominance counter and the dominant solution set are used for each solution, the order
of Pareto solutions is faster than NSGA. Therefore, the sorting algorithm defined is called the fast
dominant sorting algorithm.
Crowded Distance
The NSGA-II algorithm uses the crowded distance approach so that the solutions within the Pareto
solution set are spread and varied. The aim here is to determine the Euclidean distance between the
functional values of the solutions on a predetermined dominated surface.
The crowded distance of a solution is the distance between the related solution and neighboring
solutions on the surface on which the solution is located. It is used to estimate the perimeter of
cuboid, which is formed by using corners which are the closest neighbors, and this is called
crowded distance. For the calculation of the crowded distance, the solutions must be on the same
optimal surface. The crowded distance algorithm is as follows:
Step 1: The starting distance for all element 𝑗 on element 𝐹 is defined as zero, 𝐹 𝑑
= 0.
Step 2: For each objective function 𝑚;
The elements on the surface of 𝐹 are sorted based on the objective function 𝑚,
𝐼 = 𝑟𝑎𝑛𝑘(𝐹 , 𝑚)
The boundary value elements on the surface 𝐹 are assigned an infinite distance,
𝐼(𝑑 ) = ∞ and 𝐼(𝑑 ) = ∞
All other points are calculated,
𝐼(𝑑 ) = 𝐼(𝑑 ) +
𝐼(𝑘 + 1). 𝑚 − 𝐼(𝑘 − 1). 𝑚
𝑓
−𝑓
,
𝑘 = 2,3, … , 𝑛 − 1.
where 𝐼(𝑘 + 1). 𝑚, indicates the 𝑚. function value of element 𝑘. in set of 𝐼.
130
New population 𝑄 is generated using crossover and mutation operators for the first population
𝑃 , sorted by the fast non-dominant sorting algorithm and calculated by the crowded distance
algorithm.
Tournament Selection Operator
The selection of the crowded tournament is the process of selecting units for the match pool. The
solution is selected up to the size of the match pool. In this method, two individuals are randomly
selected. Two individuals are selected according to their dominance order and crowd distance in
the population.
This tournament process is concluded with the completion of the population size initially defined
for the next iteration.
Crossover and Mutation Operator
The crossover process is a genetic operator designed to exchange genes between parents to produce
two new chromosomes and to obtain the best properties from each. This process takes place during
the evolution period according to the user-definable crossover probability. The crossover operator
usually uses single-point crossover, two point and k point crossover, crossover for ordered lists,
and uniform crossover.
Mutations are hereditary changes that occur due to other reasons than gene change in a
chromosome. It occurs when one or more gene values on a chromosome take a different value than
the first state. In this process, the addition of new gene values to the gene pool can be achieved by
a better solution than the previously formed solution. It is also used to prevent the problem from
being stuck in local solutions in the population.
Application
Multi-objective optimization problems can be formulated as follows:
131
(1)
min 𝑓(𝑥) = [ 𝑓 (𝑥), 𝑓 (𝑥), … , 𝑓 (𝑥) ]
where 𝑥 = (𝑥 , 𝑥 , … , 𝑥 ) is the n decision variables and 𝑓(𝑥) is the m objective functions.
Reference [17] used three decision variables and five objective functions. So the model for
optimization problem:
𝑚𝑖𝑛 𝑓(𝑥) = [ 𝑓 (𝑥), 𝑓 (𝑥), 𝑓 (𝑥), 𝑓 (𝑥), 𝑓 (𝑥)]
(2)
where;
𝑓 (𝑥): surface roughness (R),
𝑓 (𝑥): tangeltial force (P1),
𝑓 (𝑥): axial force (P2),
𝑓 (𝑥): radial force (P3) and
𝑓 (𝑥): tool life (T) function.
Also 𝑥 refers to cutting velocity, 𝑥 refers to feed rate and 𝑥 refers to depth of cut variables.
Regression model for surface roughness (R):
𝑓 (𝑥) = 12.7937 − 0.03118 ∗ 𝑥[1] − 28.8786 ∗ 𝑥[2] −
2.8599 ∗ 𝑥[3] + 0.0358 ∗ 𝑥[1] ∗ 𝑥[2] +
0.00236 ∗ 𝑥[1] ∗ 𝑥[3] + 11 ∗ 𝑥[2] ∗ 𝑥[3] +
0.0000381 ∗ 𝑥[1] ∗ 𝑥[1] + 32.039 ∗ 𝑥[2] ∗ 𝑥[2] +
0.2853 ∗ 𝑥[3] ∗ 𝑥[3]
Regression model for tangeltial force (P1):
𝑓 (𝑥) = −373.0294 + 0.5308 ∗ 𝑥[1] + 788.3909 ∗ 𝑥[2] +
697.2733 ∗ 𝑥[3] − 7.2420 ∗ 𝑥[1] ∗ 𝑥[2] −
1.9860 ∗ 𝑥[1] ∗ 𝑥[3] + 235 ∗ 𝑥[2] ∗ 𝑥[3] +
0.0075 ∗ 𝑥[1] ∗ 𝑥[1] + 6659.8597 ∗ 𝑥[2] ∗ 𝑥[2] +
0.5986 ∗ 𝑥[3] ∗ 𝑥[3]
Regression model for axial force (P2):
𝑓 (𝑥) = 375.4951 − 2.971 ∗ 𝑥[1] − 360.2475 ∗ 𝑥[2] +
76.6834 ∗ 𝑥[3] + 7.9052 ∗ 𝑥[1] ∗ 𝑥[2] −
0.4 ∗ 𝑥[1] ∗ 𝑥[3] − 145 ∗ 𝑥[2] ∗ 𝑥[3] +
132
0.00398 ∗ 𝑥[1] ∗ 𝑥[1] − 1528.4596 ∗ 𝑥[2] ∗ 𝑥[2] +
66.7154 ∗ 𝑥[3] ∗ 𝑥[3]
Regression model for radial force (P3):
𝑓 (𝑥) = 239.6985 − 2.4094 ∗ 𝑥[1] + 755.0606 ∗ 𝑥[2] +
133.1862 ∗ 𝑥[3] − 0.05559 ∗ 𝑥[1] ∗ 𝑥[2] +
0.2472 ∗ 𝑥[1] ∗ 𝑥[3] − 585 ∗ 𝑥[2] ∗ 𝑥[3] +
0.0041565 ∗ 𝑥[1] ∗ 𝑥[1] + 2593.51 ∗ 𝑥[2] ∗ 𝑥[2] +
22.9351 ∗ 𝑥[3] ∗ 𝑥[3]
Regression model for tool life (T):
𝑓 (𝑥) = (𝑥[1]^0.5937) ∗ (𝑥[2]^0.4697) ∗ (𝑥[3]^0.4743)
The parameter values used for the NSGA-II method were determined by the number of variables,
the number of objective functions, the lower and upper limits of the variables for each optimization
problem. The lower and upper limits of the variables are shown in Table I. The parameters used in
the NSGA-II method are shown in Table II.
TABLE VIII.
LOWER LIMIT AND UPPER LIMIT VALUES OF CUTTING PARAMETERS
Parameters
Lower Bound
Upper Bound
cutting velocity (𝒙𝟏 )
142
265
feed rate (𝒙𝟐 )
0.15
0.25
depth of cut (𝒙𝟑 )
1
2
TABLE IX.
PARAMETERS USED IN NSGA-II METHOD
Parameters
Value
Population size
100
Generation size
100
Crossover rate
0.7
Mutation rate
0.2
133
Pareto solution set was obtained for optimal solution of these functions. The results obtained from
the studies conducted by [1] and [17] were compared with the most appropriate results in the
NSGA-II method set and shown in Table III.
COMPARISON OF OPTIMIZATION RESULTS
TABLE X.
𝑥
𝑥
𝑥
R
P1
P2
P3
T
[17]
235 0.15 1.00 4.85 445.13 115.39 199.14 40.34
[1]
225 0.15 1.15 4.90 479.09 135.15 224.28 38.74
NSGA-II 235 0.15 1.00 4.84 445.97 114.71 198.96 40.26
According to the results of the analysis, the optimal results of the objective functions in the study
by [17]: 4.85 for the surface roughness (R), 445.13 for the tangeltial force (P1), 115.39 for the
axial force (P2), 199.14 for the radial force (P3) and 40.34 for tool life (T) was found. In case the
variables have the same value, the objective function values obtained by NSGA-II method are:
4.84 for the surface roughness (R), 445.97 for the tangeltial force (P1), 114.71 for the axial force
(P2), 198.96 for the radial force (P3) and 40.26 for tool life (T) was found. Better results were
obtained with the objective function values obtained by NSGA-II method. In the study conducted
by [1], although the results obtained by the genetic algorithm method of the optimization problem
are close to the results obtained with the other two methods, it can not be said that it is better.
Results
In this study, NSGA-II approach which is effective in heuristic methods has been used for the
solution of multi-objective optimization problems. The surface roughness (R), tangential force
(P1), axial force (P2), radial force (P3) and tool life (T) functions are optimized for estimation of
the cutting parameters effecting the turning of hard materials as a multi-objective optimization
problem. In addition, the results were compared with the results of the study by [1] and [17] and
it was shown that the results obtained with the NSGA-II method were better.
According to these results, the solutions obtained by using NSGA-II method have been successful
in multi-objective optimization problems. When classical methods for optimization problems with
134
multiple objective functions are insufficient, an efficient method NSGA-II can be used in heuristic
methods. In addition, decision makers from the optimal solutions can choose the most suitable
solution according to their importance in the objective functions.
REFERENCES
[1] O. Manav, And S. Chinchanikar, “Multi-Objective Optimization of Hard Turning: A Genetic
Algorithm Approach” Materials Today: Proceedings, Vol. 5, No. 5, Pp. 12240-12248, 2018.
[2] R. C. Peralta, A. Forghani, And H. Fayad, “Multiobjective Genetic Algorithm Conjunctive Use
Optimization For Production, Cost, And Energy With Dynamic Return Flow” Journal of Hydrology, Vol.
511, Pp. 776-785, 2014.
[3] Ç. Elmas, “Artificial Intelligence Applications, Artificial Neural Networks - Fuzzy Logic Genetic Algorithm”, 2nd Ed., Ankara: Seçkin, 2011.
[4] D. E. Goldberg, “Genetic Algorithms İn Search, Optimization, and Machine Learning”, 1st Ed.,
Usa: Addison-Wesley, 1989.
[5] S. Kirkpatrick, C. D. Gelatt, And M. P. Vecchi, “Optimization By Simulated Annealing”
Science, Vol. 220, No. 4598, Pp. 671-680, 1983.
[6] D. Kalyanmoy, “Multi Objective Optimization Using Evolutionary Algorithms” 1st Ed., New
York: John Wiley & Sons, Pp. 124, 2001.
[7] C. M. Fonesca, And P. J. Fleming, “Genetic Algorithms For Multiobjective Optimization:
Formulation, Discussion And Generalization” in Proc Of The Fifth Int. Conf. On Genetic Algorithms, Pp.
415-423, July 1993.
[8] J. Horn, N. Nafploitis, And D. E. Goldberg, “A Niched Pareto Genetic Algorithm For
Multiobjective Optimization” Nj: Ieee Press, Pp. 82-87, 1994 [Proceedings Of The First Ieee Conference
On Evolutionary Computation].
[9] N. Srinivas, And K. Deb, “Muiltiobjective Optimization Using Nondominated Sorting in
Genetic Algorithms” Evolutionary Computation, Vol. 2, No. 3, Pp. 221-48, 1995.
[10] E. Zitzler, And L. Thiele, “Multiobjective Optimization Using Evolutionary Algorithms-A
Comparative Case Study” International Conference On Parallel Problem Solving From Nature, Berlin,
Germany: Springer, Pp. 292-301, 1998.
135
[11] B. Wang, Y. Liang, T. Zheng, M. Yuan, And H. Zhang, “Multiobjective Site Selection
Optimization of The Gas-Gathering Station Using Nsga-Iı” Process Safety And Environmental Protection,
Vol. 119, Pp. 350-359, 2018.
[12] S. M. Wang, S. Ma, And W. Y. Duan, “Seakeeping Optimization of Trimaran Outrigger
Layout Based On Nsga-Iı” Applied Ocean Research, Vol. 78, Pp. 110-122, 2018.
[13] Y. Yang, L. Cao, C. Wang, Q. Zhou, And P. Jiang, “Multi-Objective Process Parameters
Optimization Of Hot-Wire Laser Welding Using Ensemble Of Metamodels And Nsga-Iı” Robotics And
Computerıntegrated Manufacturing, Vol. 53, Pp. 141-152, 2018.
[14] T. Vo-Duy, D. Duong-Gia, V. Ho-Huu, H. C. Vu-Do, And T. Nguyenthoi, “Multi-Objective
Optimization of Laminated Composite Beam Structures Using Nsga-Iı Algorithm” Composite Structures,
Vol. 168, Pp. 498-509, 2017.
[15] N. Alikar, S. M. Mousavi, R. A. R. Ghazilla, M. Tavana, And E. U. Olugu, “Application Of
The Nsga-Iı Algorithm To A Multi-Period Inventory-Redundancy Allocation Problem İn A Series-Parallel
System” Reliability Engineering & System Safety, Vol. 160, Pp. 1-10, 2017.
[16] K. Deb, A. Pratap, S. Agarwal, And T. A. M. T. Meyarivan, “A Fast And Elitist Multiobjective
Genetic Algorithm: Nsga-II” IEEE Transactions On Evolutionary Computation, Vol. 6, No. 2, Pp.182-197,
2002.
[17] S. Chinchanikar, And S. K. Choudhury, “Effect of Work Material Hardness And Cutting
Parameters on Performance Of Coated Carbide Tool When Turning Hardened Steel: An Optimization
Approach” Measurement, Vol. 46, No. 4, Pp. 1572-1584, 2013.
136
CO2 Emission and Energy Consumption for Different Climate and Building Materials
Salih HİMMETOĞLU, Yılmaz DELİCE, Emel KIZILKAYA AYDOĞAN
Department of Industrial Engineering, Erciyes University, TURKEY
Abstract: With the development of technology from past to present, the types, properties and
product range of the materials used in the buildings are quite developed. Therefore, the effects of
climate, environmental conditions and energy consumption cannot be ignored for selecting these
materials used in the buildings. Usage of materials with the same characteristics for buildings to
be built on different climate may lead to adverse effects about energy-saving and green gasses.
Furthermore, the use of the same materials may not be a proper approach even in buildings with a
different purpose. In this study, forecasting of energy consumption and CO2 emission is analyzed
by utilizing artificial neural network structure according to different climate criteria and material
characteristics for public buildings built in recent years. The Effect levels to energy consumption
and CO2 emission of the building materials and the climate criteria are determined for buildings
serving the same using purpose in different climate characteristics. For the study, different pilot
regions where the public buildings are located are chosen according to climatic characteristics and
five different building materials used in these public buildings are taken into account. When the
results compare according to CO2 emission and energy consumption, it was observed that the
conditions which obtain most efficient results are different.
Keywords: Artificial Neural Network, Energy Consumption, CO2 Emission, Data Mining,
Forecasting, Public Buildings
REFERENCES
[1] M. Bayraktar, ―A Methodology for Energy Optimization of Buildings Considering
Simultaneously Building Envelope HVAC and Renewable System Parameters, Ph. D. Thesis. 2015.
[2] M. Tekin, Production Management, 6th ed., vol. 1. Konya: Günay Ofset, 2009.
137
[3] R. M. Kirby, "A Comparison of Short And Medium Range Statistical Forecasting Methods"
Management Science Vol. 13 Pp. 4, B-202, 1966.
[4] R. L. Carlson And M. Umble, "Statistical Demand Functions For Automobiles and Their Use
For Forecasting İn An Energy Crisis," The Journal Of Business, Pp. 193 – 204, 1980.
[5] S. L. Zhou, T. A. Mcmahon, A. Walton And J. Lewis, ―Forecasting Operational Demand For
An Urban Water Supply Zone, Journal Of Hydrology, Vol. 259(1-4), Pp. 189-202, 2002.
[6] E. E. Cahow, ―Forecast Of Demand For Chronic Care Nursing Home Services: 2005-2025,
Brandeis University, Waltham Massachusetts, USA, Ph. D. Thesis, 2003.
[7] K. Matuyama, T. Sumita And D. Wakayama, ―Periodic Forecast and Feedback To Maintain
Target Inventory Level, International Journal Of Production Economics, Vol. 118(1), Pp. 298-304, 2009.
[8] R. Fildes, P. Goodwin, M. Lawrence And K. Nikolopoulos, ―Effective Forecasting and
Judgmental Adjustments: An Empirical Evaluation And Strategies For İmprovement in Supply-Chain
Planning, International Journal Of Forecasting, Vol. 25, Pp. 3–23, 2009.
[9] X, Xu, Y. Qi And Z. Hua, ―Forecasting Demand of Commodities After Natural Disasters,
Expert Systems With Applications, Vol. 37, Pp. 4313-4317, 2010.
[10] W. E. Griffiths, L. S. Newton And C. J. O’Donnell, ―Predictive Densities For Models With
Stochastic Regressors And Inequality Constraints: Forecasting Local-Area Wheat Yield, International
Journal of Forecasting, Vol. 26(2), Pp. 397-412, 2010.
[11] M. Ghodsi, ―A Brief Review Of Recent Data Mining Applications İn The Energy İndustry,
International Journal Of Energy And Statistics, Vol. 2(01), Pp. 49 - 57, 2014.
[12] M. W. Gardner And S. R. Dorling, ―Artificial Neural Networks (The Multilayer
Perceptron)—A Review of Applications in The Atmospheric Sciences, Atmospheric Environment, Vol.
32(14-15), Pp. 2627-2636, 1998.
[13] X. Sun, D. K. Gauri And S. Webster, ―Forecasting For Cruise Line Revenue Management,
Journal of Revenue and Pricing Management, Doi:10.1057/Rpm.2009.55, 2010.
[14] S. Chung, ―Demand Modeling And Analysis For The Management of Underground
Infrastructure Systems, Purdue University, USA, Ph. D. Thesis, 2001.
[15] H. Shuai, G. Qingwu And J. Wu, ―A New Multi-Method Combination Forecasting Model
For ESDD Predicting, IEEE Asia-Pacific Power And Energy Engineering Conference Pp. 1 – 4, March
2010.
138
[16] M. R. Devi And R. Manonmani, " Electricity Forecasting Using Data Mining Techniques in
Tamil Nadu and Other Countries - A Survey," International Journal of Emerging Trends in Engineering
And Development, Vol. 6(2), Pp. 295–302, 2012.
[17] L. Fugon, J. Juban And G. Kariniotakis, ―Data Mining For Wind Power Forecasting,
European Wind Energy Conference & Exhibition EWEC 2008, Brussels, Belgium, Pp. 1–6, 2008.
[18] S. R. Mohandes, X. Zhang And A. Mahdiyar, ―A Comprehensive Review on the Application
of Artificial Neural Networks in Building Energy Analysis. Neurocomputing, Vol. 340, Pp. 55-75, 2019.
[19] X. Cipriano, J. Carbonell And J. Cipriano, "Monitoring And Modeling Energy Efficiency of
Municipal Public Buildings: Case Study in Catalonia Region," International Journal Of Sustainable Energy,
Vol. 28(1-3), Pp. 3-18, 2009.
[20] D. Hawkins, S. M. Hong, R. Raslan, D. Mumovic, And S. Hanna, ―Determinants of Energy
Use in UK Higher Education Buildings Using Statistical And Artificial Neural Network Methods,
International Journal of Sustainable Built Environment, Vol. 1(1), Pp. 50-63, 2012.
[21] H. Son, S. Lee, C. Kim, "An Artificial Neural Network-Based Prediction Of GovernmentOwned Building Energy Consumption With Design Variables," Pro-Ceedings Of The ICSDEC 2012
Developing The Frontiers Of Sustainable Design Engineering Construction, Pp. 1–10, 2013.
[22] C. Buratti, M. Barbanera And D. Palladino. "An Original Tool For Checking Energy
Performance and Certification Of Buildings By Means Of Artificial Neural Networks," Applied Energy,
Vol. 120, Pp. 125- 132, 2014.
[23] S. Mishra And V. K. Singh, ―Monthly Energy Consumption Forecasting Based on Windowed
Momentum Neural Network, IFAC-Papersonline, 2015, Vol. 48(30) Pp. 433-438, 2015.
[24] Q. Dong, K. Xing And H. Zhang, ―Artificial Neural Network For Assessment of Energy
Consumption and Cost For Cross Laminated Timber Office Building İn Severe Cold Regions,
Sustainability, Vol. 10(1), Pp. 84, 2017.
[25] C. Deb, S. E. Lee And M. Santamouris, ―Using Artificial Neural Networks to Assess HVAC
Related Energy Saving İn Retrofitted Office Buildings, Solar Energy, Vol. 163, Pp. 32-44, 2018.
139
Bitcoin Price Forecasting with Multivariate Long Short Term Memory (LSTM) Deep
Learning Method
Ali Osman ÇIBIKDİKEN1, Ebru Şeyma KARAKOYUN2
1
Department of Computer Engineering, Necmettin Erbakan University,
2
Department of Industrial Engineering, Karabük University, TURKEY
Abstract: Long Short Term Memory (LSTM) is one of the deepest learning methods capable of
learning along a chain. The method has a chain of modules able to repeating information and
transferring it to the next module. Due to this feature, it is a convenient method for data sets
consisting of time-dependent information such as finance. Bitcoin, using blockchain technology,
has become one of the most popular cryptocurrencies today. Bitcoin data is a time series. In this
study, price estimation model is proposed by using Long-Short Term Memory method for a Bitcoin
price estimation for multivariate time series consisting of opening price, closing price, highest
price, lowest price, Bitcoin volume, Purchasing volüme and weighted price variables. In addition,
the application has been developed in Python programming language.
REFERENCES
[1] S. Hochreiter, J. Schmidhuber, “Long Short-Term Memory”, Neural Computation, 9(8):17351780, 1997.
[2] S. Nakamoto, “Bitcoin: A Peer-To-Peer Electronic Cash System”, 2008.
[3] A. Navon, Y. Keller, “Financial Time Series Prediction Using Deep Learning”,
Https://Arxiv.Org/Abs/1711.04174, 2017.
[4] F. Thomas, K. Christopher, “Deep Learning With Long Short-Term Memory Networks For
Financial Market Predictions”, Fau Discussion Papers İn Economics, No. 11/2017, 2017.
[5] W. W. S. Wei, Time Series Analysis Univariate and Multivariate Methods, Pearson Education,
2006.
[6] F. Chollet, Keras, Https://Github.Com/Fchollet/Keras, 2016.
140
Covering-Based Generalized IF-Rough Set Models For A Selecting HVAC System
Salih HİMMETOĞLU, Emel KIZILKAYA AYDOĞAN, Yılmaz DELİCE
Department of Industrial Engineering, Erciyes University, TURKEY
Abstract: In today's urban life, most of the people's time is spent at home or work office. For this
reason, design and selecting of building materials that control heating (H), ventilation (V) and air
conditioning (AC) are essential. The building materials called, in short HVAC, must
simultaneously provide high comfort, low cost and high energy productivity. Furthermore, HVAC
must be appropriately designed to prevent adverse effects on the environment and climate. In this
study, nine different HVAC systems were examined according to nine different criteria over cost,
pollution, comfort and energy which are considered as four main factors in the selection of HVAC
systems. Since some of the criteria for HVAC systems are described as linguistic, it is not possible
to evaluate the systems with traditional methods using crisp values. Therefore, we propose
generalized intuitionistic fuzzy (IF) - rough set models which are a new and flexible method. IFneighborhoods are formed by using IF-implicator and IF-t norms, and upper and lower
approximations in rough set theory are calculated according to the neighborhoods. Covering-based
generalized IF rough set models are generated by using the approximations and IF-TOPSIS
method. According to the obtained results, we can see that the proposed method is an appropriate
decision-making method which considers the uncertainties in the linguistic expressions for
selecting the most suitable HVAC system.
Keywords: HVAC, IF-Covering, Rough Set Theory, Fuzzy Sets, Approximations
REFERENCES
[1] M. Bayraktar, “A Methodology for Energy Optimization of Buildings Considering
Simultaneously Building Envelope HVAC and Renewable System Parameters,” Ph. D. Thesis, 2015.
[2] E. Burak, F. E. Boran, and M. Kurt, “Ergonomic Product Concept Selection Using Intuitionistic
Fuzzy TOPSIS Method,” Journal of Engineering Sciences and Design, vol. 3(3), pp. 433-440, 2015.
141
[3] L. Zhang, J. Zhan, And Z. Xu, “Covering-Based Generalized If Rough Sets With Applications
To Multi-Attribute Decision-Making,” Information Sciences, Vol. 478, Pp. 275-302, 2019.
[4] J. S. Dodgson, M. Spackman, A. Pearman, And L. D. Phillips, Multicriteria Analysis: A Manual,
Department For Communities And Local Government, London, 2009.
[5] J. P. Huang, K. L. Poh, And B. W. Ang, “Decision Analysis in Energy And Environmental
Modeling,” Energy, Vol. 20(9), Pp. 843-855, 1995.
[6] J. D. Balcomb, And A. Curtner, “Multi-Criteria Decision-Making Process For Buildings. In
Collection Of Technical Papers.” Ieee 35th Intersociety Energy Conversion Engineering Conference And
Exhibit, Vol 1, Pp. 528-535, 2000.
[7] S. De Wit, And G. Augenbroe, “Analysis Of Uncertainty İn Building Design Evaluations And
İts İmplications,” Energy And Buildings, Vol. 34(9), Pp. 951-958, 2002.
[8] C. J. Hopfe, Uncertainty And Sensitivity Analysis İn Building Performance Simulation For
Decision Support And Design Optimization, Eindhoven University, Ph. D. Thesis, 2009.
[9] S. H. Kim, And G. Augenbroe, Ventilation Operation in Hospital İsolation Room: A MultiCriterion Assessment Considering Organizational Behavior in Building Simulation, Pp. 27-30, July 2009.
[10] Y. J. Kim, K. U. Ahn, And C. S. Park, “Decision Making of Hvac System Using Bayesian
Markov Chain Monte Carlo Method,” Energy And Buildings, Vol. 72, Pp. 112-121, 2014
[11] P. Huang, G. Huang, And Y. Wang, “Hvac System Design Under Peak Load Prediction
Uncertainty Using Multiple-Criterion Decisionmaking Technique," Energy And Buildings, Vol. 91, Pp. 2636, 2015.
[12] K. T. Atanassov, Intuitionistic Fuzzy Sets. In Intuitionistic Fuzzy Sets, Physica, Heidelberg,
1999, Pp. 1-137.
[13] Z. H. Xu, “Intuitionistic Fuzzy Aggregation Operators,” IEEE Transactions on Fuzzy Systems,
Vol. 15(6), Pp. 1179-1187, 2007.
142
Determination of Production Defects in Iron and Steel Sector by Data Mining
İsmail Burak AKINCI, Filiz ERSÖZ
Industrial Engineering Department, Karabük University, TURKEY
Abstract: The studies related to the production industry are limited in the world and in our country.
Especially in iron and steel sector, quality levels of different types of products need to be
monitored. Iron and steel products obtained from the studies have prolonged their use and price
and sales superiority has been achieved. At the same time, the market value of the products
increases and there is a minimum loss of product. Therefore, studies in this field should be focused
on. On the basis of quality, instead of debugging errors is the approach of not making mistakes.
Instead of using your earnings as a philosophy, we should adopt an understanding of gaining from
our losses. Understanding the importance of quality work and improvements, the primary purpose
of enterprises is to support quality production by preventing or reducing errors in production. Data
mining has started to be used effectively in enterprises. Data mining involves the process of
selecting, organizing and modeling the most necessary data for business executives. At this point,
it is possible to define data mining as a set of techniques and concepts that produce new
information for decision-making processes. In this study, firstly the VM process is defined and
then the VM studies which are selected from the literature covering 2010-2018 and applied to
certain quality improvement problems in the manufacturing sector are evaluated. The definition of
process and product quality, estimation of quality, classification of quality and optimization of
quality parameters are discussed. In addition, the application of decision trees, one of the most
widely used and effective VM techniques, in order to determine the variables and levels that cause
production errors in an industrial organization is also included.
Keywords: Production, Manufacturing Defect, Data Mining
143
REFERENCES
[1] Liu, H.T., Tsai, Y.L. “A Fuzzy Risk Assessment Approach For Occupational Hazards in The
Construction Industry”, Safety Science. 50,1067-1078 (2012).
[2] Dünya Çelik Derneği [Online]. Erişim adresi: http://celik.org.tr/2017-yilinda-dunya-ham-celikuretimi-5-3-oraninda-artti/ Erişim tarihi: 3 Kasım 2018.
[3] Türkiye Çelik Üreticileri Derneği [Online]. Erişim adresi: http://celik.org.tr/harita/ Erişim
tarihi: 5 Kasım 2018.
[4] Güner, M., Akman, Ü., Yücel, Ö. “Improving The Man Shirt Production Processusing The Six
Sigma Method”, Tekstil ve Konfeksiyon, 20 (1), 75-82 (2010).
[5] Uyran, B. “Toplam Kalite Yönetimi”, Mevzuat Dergisi, 55 (2002).
[6] Çevik, O., Aran, G. Kalite İyileştirme Sürecinde Hata Türü Etkileri Analizi (Fmea) Ve Piston
Üretiminde Bir Uygulama. Sosyal Ekonomik Araştırmalar Dergisi, 8 (16), 241-265 (2009).
[7] Ersöz F., “Veri Madenciliği Teknikleri ve Uygulamaları” Ankara, (2015).
[8] Kayaalp, F., Başarslan, M., “Open Source Data Mining Programs: A Case Study on R”, 6, 4245 (2018).
[9] Savaş, S., Topaloğlu, N., Yılmaz, M., “Veri Madenciliği ve Türkiye’deki Uygulama Örnekleri”
(2012).
[10] Sumathi, S., Sivanandam, S.N., “Introductionto Data Miningand Its Applications” (2006).
[11]
Veri
Madenciliği
ve
İşletmelerde
Kullanım
Alanları
[Online].
Erişim
adresi:
https://dralabay.wordpress.com/2014/03/07/veri-madenciligi-ve-isletmelerde-uygulama-alanlari/Erişim
tarihi: 5Kasım 2018.
[12] Veri Madenciliği Aşamaları [Online]. Erişim adresi:https://mevlutcanvar.com.tr/verimadenciligi-asamalari/309Erişim tarihi: 5Kasım 2018.
[13]
Veri
Madenciliği
Yöntemleri
[Online].
adresi:https://emraharslanbm.wordpress.com/2012/07/17/veri-madenciligi-yontemleri/Erişim
Kasım 2018.
[14] Yaralıoğlu, K., “Dokuz Eylül Üniversitesi Veri Madenciliği Ders Notu”.
144
Erişim
tarihi:
5
[15] ISO Nedir [Online]. Erişim adresi: https://www.muhendisbeyinler.net/iso-nedir/ Erişim tarihi:
5Ocak 2019.
[16] ISO Standartları [Online]. Erişim adresi: https://www.iso.org/obp/ui/#iso:std:iso:9001:ed5:v1:enErişim tarihi: 5Ocak 2019.
[17] ISO 9001 Genel Bilgi [Online]. Erişim adresi: https://www.iso.org/iso-9001-qualitymanagement.html Erişim tarihi: 5Ocak 2019.
145
Connected Employee Platform and A Case Study in A Global Company
Batuhan Burak ERSÖZLÜ, Ufuk CEBECİ, Şahen TOKATLIOĞLU
Department of Industrial Engineering, Istanbul Technical University, TURKEY
Abstract: In the modern era, enterprises are facing a variety of difficulties because of today’s
emerging technologies. One of the numerous difficulties for a business is to fulfill its employees
in order to adapt to the consistently changing business processes and to make progress and stay in
competition. In order to build proficiency, viability, efficiency and occupation responsibility of
employee, the business must fulfill the requirements of its employee by giving great working
conditions. The target of this paper is to dissect the effect of workplace on employee work
fulfillment. This paper may profit society by urging individuals to contribute more to their
occupations and may help them in their daily work life. Consequently, it is fundamental for an
association to support their employee to snap down for accomplishing the hierarchical objectives
and goals. The investigation and the item are changing the advanced endeavor, expanding
representative commitment over the whole workforce, including forefront, field, remote and
outside laborers to enhance execution, efficiency and unwaveringness. Our stage is making
ground-breaking employee encounters, where every single representative feels some portion of an
option that is more noteworthy than themselves, are glad to be a piece of your image and effectively
advance the positive characteristics of the association.
Keywords: Employee Platforms, Employee Satisfaction, Employee Experience, Case Study,
Mobile Application, User Experience, Software, Connected Workspace, Human Resources,
İnformation Technologies, İnnovation, Visualized Applications
REFERENCES
[1] Forrester Research: Build Mobile Apps That Drive Engagement by Jeffrey S. Hammond,
Michael Facemire, and Rowan Curran with Christopher Mines and Eric Wheeler, February 23, 2015
[2] Forrester Research: Trends 2015: The Future Of Customer Service by Kate Leggett with
Stephen Powers, Ian Jacobs, and Arelai Ephraim, March 2, 2015
146
[3] Predictions on the Future of the Employee Experience. (2019). Retrieved from
https://www.hrtechnologist.com/articles/employeeengagement/5-predictions-on-the-future-of-theemployee-experience/
[4]
The
Connected
Employee
Experience.
(2019).
Retrieved
from
https://
www.capgemini.com/service/technology-operations/infrastructureservices/connected-employeeexperience/
[5] Your own branded employee app for Internal Communications, Engagement and HR. (2019).
Retrieved from https://staffbase.com/en/
[6] Trends Shaping the Future of Employee Experience in 2019. (2019). Retrieved from
https://www.itagroup.com/insights/trends-shapingfuture-of-employee-experience
[7] Baca, B., & Cassidy, A. (1999, September). Intranet Development And Design That Works. In
Proceedings Of The Human Factors And Ergonomics Society Annual Meeting (Vol. 43, No. 13, pp. 777781). Sage CA: Los Angeles, CA: SAGE Publications.
[8] Bersin, J. Becoming irresistible: A New Model For Employee Engagement, Deloitte Review
16, January 26, 2015, Retrieved from https://www2.deloitte.com/insights/us/en/deloitte-review/issue-16/
employee-engagement-strategies.html .
[9] Bottazzo, V. (2005). Intranet: A Medium of Internal Communication and Training. Information
Services & Use, 25(2), 77-85.
[10] Buniyamin, N., & Barber, K. D. (2004). The intranet: A Platform For Knowledge Management
System Based On Knowledge Mapping. International Journal of Technology Management, 28(7-8), 729746.
[11] Cole, C. (2019). BMC's 'Connected' Employee Experience Boots Digital Workplace. [online]
DiversityQ.
Available
at:
https://diversityq.com/
bmcs-connected-employee-experience-boots-
digitalworkplace-1004791/ [Accessed 4 Feb. 2019].
[12] Damsgaard, J., & Scheepers, R. (2001). Harnessing Intranet Technology For Organisational
Knowledge Creation. Australasian Journal of Information Systems, 9(1).
[13] Damsgaard, J., & Scheepers, R. (2001). Using Intranet Technology To Foster Organizational
Knowledge Creation. ECIS 2001 Proceedings, 32.
147
[14] Dukes, E. (2019). The Business Case for Employee Experience Software. [online]
Iofficecorp.com.
Available
at:
https://
www.iofficecorp.com/blog/the-business-case-for-
employeeexperience-software [Accessed 4 Feb. 2019].
[15] Grover, R. (2019). How to Boost Employee Engagement with Awesome Internal
Communication.
Retrieved
from
https://
insights.staffbase.com/blog/18-ideas-to-improve-
internalcommunication-in-2018-and-boost-your-employee-engagement.
[16] Horton, R., Buck, T., Waterson, P. (2001). Explaining Intranet Use With The Technology
Acceptance Model. Journal of Information Technology. 16, 237.
[17] Intranet Software That Connects Your Organization | Interact Software. (2019). Retrieved
from https://www.interact-intranet.com
[18] Kaplan, R. S., & Norton, D. P. (1995). Putting the Balanced Scorecard. Performance
measurement, management, and appraisal sourcebook, 66.
[19] Lockley, S. (2019). How Employee Experience Is the Way to a More Engaging Workplace.
Retrieved from https://insights.staffbase.com/ blog/the-anatomy-of-an-employee-experience-refiningtodaysworkplace
[20] MacCormick J.S., Dery K., Kolb D.G. (2012). Engaged or Just Connect? Smartphones And
Employee Engagement. Organizational Dynamics, 41, 194.
[21] Mahapatra, R., & Lai, V. (1998). Intranet-based Training Facilities ERP System
Implementation: A Case Study. AMCIS 1998 Proceedings, 366.
[22] Murgolo-Poore, M. E., Pitt, L. F., & Ewing, M. T. (2002). Intranet Effectiveness: a Public
Relations Paper-And-Pencil Checklist. Public Relations Review, 28(1), 113-123.
[23] Mphidi, H., & Snyman, R. (2004). The Utilisation of An Intranet As A Knowledge
Management Tool in Academic Libraries. The Electronic Library, 22(5), 393-400.
[24] The employee experience: Helping People Get Excited To Do Their Best At Work. (2019).
Retrieved from https://www.pwc.com/us/en/library/ workforce-of-the-future/employee-experience.html
[25] Plaskoff, J. (2017). Employee Experience: The New Human Resource Management Approach.
Strategic HR Review,16 (3), 136-141.
[26] Scott, J.E. (1998). Organizational Knowledge And The Intranet. Decision Support Systems,
23(1), 3–17.
148
[27] Scarbrough, H. (1999) Feature Knowledge Management: System Error. People Management,
April, 68–74.
[28]
The
Connected
Employee
Experience.
(February,
2014).
Retrieved
from
https://www.pwc.com/mt/en/publications/assets/hrs-pwctechnology-connected-employee-experience.pdf
[29] Sollenberger, D., Darpel, D., Talevski, S., Shumaker, L., Meyer, K., Journey, J., ... & Gimbert,
N. (2002). U.S. Patent Application No. 09/833,433.
[30] Ruppel, C. P., & Harrington, S. J. (2001). Sharing Knowledge Through İntranets: A Study of
Organizational Culture And İntranet İmplementation. IEEE transactions on professional communication,
44(1), 37-52.
[31] Urbach, N., Smolnik, S., & Riempp, G. (2009). A Conceptual Model For Measuring The
Effectiveness of Employee Portals. AMCIS 2009 Proceedings, 589.
[32] Urbach, N., Smolnik, S., & Riempp, G. (2010). Improving the Success of Employee Portals:
A Causal and Performance-Based Analysis. In ECIS (p. 54).
[33] Tojib, D. R., Sugianto, L. F., & Sendjaya, S. (2008). User Satisfaction With Business-ToEmployee Portals: Conceptualization And Scale Development. European Journal of Information Systems,
17(6), 649-667.
[34]
What
is
iBeacon?
A
Guide
to
iBeacons.
(2019).
Retrieved
from
http://
www.ibeacon.com/what-is-ibeacon-a-guide-to-beacons/
[35] Bluetooth | Definition of Bluetooth in English by Oxford Dictionaries. (2019). Retrieved from
https://en.oxforddictionaries.com/definition/ bluetooth
[36]
Android
(operating
system).
(2019).
Retrieved
from
https://
en.wikipedia.org/wiki/Android_(operating_system)
[37] IOS. (2019). Retrieved from https://en.wikipedia.org/wiki/IOS 38. GPS (Global Positioning
System) Definition. (2019). Retrieved from https://techterms.com/definition/gps
149
Investigation of the Effects of Normal Distribution or Nonnormal of Data on Machine
Capability Analysis
Erkan IŞIĞIÇOK, Gözde TÜRK
Department of Econometrics, Uludağ University, TURKEY
All processes in the business world have two fundamental disease, including deviation and
variability from the average (target). One of the statistical process control graphs used for
quantitative variables is to keep the average and the other to control variability. Apart from the
normal distribution or nonnormal of quantitative data, the average and variability are controlled or
not, and then the capability of process or machine is checked. The desired outcome is in addition
to the normal distribution of data, the process is under control and capable. On the other hand,
capability analysis is defined as the machine capability analysis when it is performed for the
machine, while the process capability analysis takes its name when it is done for the process. In
this study, machine capability analysis has been applied.
The aim of this article is to investigate the effects of the normal distribution or nonnormal of data
on machine capability analysis. For this purpose, data on the lengths measured by surface of the
shock absorber body pipe cut by a CNC machine in a company in the automative industry were
used. In the study, 50 observations values were used, and the lower specification limit was 124.5
and upper specification limit was 125.5, and the CNC (pipe cutting) machine was sufficient or not.
The analysis first started with the implementation of the normality test and the data was not
distributed normally. It is concluded that assuming this data, which does not have normal
distribution, is normally distributed, and the machine is under control and is also sufficient with
the I-MR control charts in the Minitab program. The same analysis was applied with the nonnormal
command under the assumption that the data was nor normally distributed, and even in this case
the machine was sufficient. In addition, the data that does not have normal distribution has been
transformed into normality, and I-MR control charts and machine are under control and also
sufficient. According to the findings, the average and specification limits of the values of I-MR
control charts are the same and the machine capability results are different. In this study, these
150
similarities and differences were examined comparatively. Let us add it right away; these findings
are specific to the machine and cut pipes we take into consideration and should not be generalized.
Keywords: Normal Distribution, The Data Does Not Have A Normal Distribution, Control Charts
for Measurements, Under Control, Machine Capability Analysis.
151
Portfolio Selection with the Possibilistic Mean – Variance Model: An Application on the
Borsa Istanbul
Furkan GÖKTAŞ, Süleyman DÜNDAR
Department of Business Administration, Karabük University, TURKEY
Abstract: The possibilistic mean – variance (MV) model enables the practitioners to incorporate
the expert knowledge and robust statistics into the portfolio selection. Hence, it is a considerable
alternative in decision making under uncertainty. In this study, we will examine the possibilistic
mean – variance model theoretically under the assumption that the possibility distributions of the
asset returns are given with the triangular fuzzy numbers. Here, the triangular fuzzy numbers will
be determined based on the box plots. According to this, the possibilistic mean depends on the
data set’s median, interquartile range and skew. Furthermore, the possibilistic variance depends
only on the data set’s interquartile range. Then, we will illustrate this model based on the weekly
returns of ten sector indices in 2017. Moreover, we will compare the risk adjusted performance
and profitability of the possibilistic MV model and Markowitz’s traditional MV model where the
trading and testing periods cover the complete year of 2017 and 2018 respectively.
Keywords: Portfolio Selection, Decision Making Under Uncertainty, Triangular Fuzzy Numbers,
Box Plot, Robust Statistics, Linear Programming
152
Building Digital Assistant (ChatBot) with SAP Conversational Artificial Intelligence
Metehan KOCAOĞLU1, Kağan ÖZDEMİR1, Alper KARABULUT1,
Rüştü Orkun KORKMAZ1, Ufuk CEBECİ2
1
Solvia Digital Solutions, 2Department of Industrial Engineering,
Istanbul Technical University, TURKEY
Abstract: Human-Computer Speech is gaining momentum as a technique of computer
interaction. A chatbot is a software, which can “chat" with a human user in natural language such
as English. Conversational artificial intelligence technology (CAI) enables learners to engage in
spoken conversations with the non-player characters. In this study, SAP's digital assistant can be
used to run business processes using SAP Conversational Artificial Intelligence. The benefits of
the digital assistant (chatBot) will be discussed. An example of the request for personnel’s leave
request, which is one of the areas of use, will be explained.
Keywords: Chatbot, Digital Asistant, SAP CAI, Conversational Artificial Intelligence, Leave
Request.
Introduction
Nowadays, many innovations have started to be created and used with the digitalizing world. In
addition, companies are trying to integrate innovations with their own systems in order to manage
their processes quickly. It can be given 11 examples about chatbot. WeChat, one of them, was
created by Chinese holding company Tencent in 2013. The product was created by a special
projects team within Tencent (who also owns the dominant desktop messaging software in China,
QQ) under the mandate of creating a completely new mobile-first messaging experience for the
Chinese market [3]. In this study, digital assistant (ChatBot) application which can be created in
SAP by using artificial intelligence was performed. The aim of ChatBot is to be a digital assistant.
Today, time saving is a must-have. Humanity is looking for ways to make standardized tasks
simpler, faster and more useful throughout the day. Many new applications are emerging new
153
products with this quest. SAP CAI is an innovative and self-enhancing bot creation platform. SAP
CAI provides some significant gains [1].
Saving Time
One of the biggest benefits of using chatBot in your business is to save time. When used
on the website, it can provide quick and automatic answers to most questions and prevents
customers from waiting for a day or longer to get a response as in the past. This allows an enterprise
to serve more people while increasing productivity and reducing costs.
Saving Money
The use of Chatbot may be cheaper than employing more workers. Thanks to algorithms
to be created for frequently asked questions, you can meet the needs of your staff through chatBot
without having to find a person.
Customer Satisfaction
Another benefit of using the conversations in your business is that they provide more
customer satisfaction. Chat robots do not work for 8 hours and do not need sleep, this means they
are always available. Customers using the website in the evening can ask questions and get
immediate answers. This could increase your profits. Frustrated customers who can't get a quick
reply may not leave your website and come back. Chat boots can eliminate this scenario.
Increasing Customer Base
Another benefit of using the conversations in your business is that they can help you reach
more people to increase your customer base. Since chatBots can be used in many applications, it
can be used to help the business grow.
Reducing Errors
Unfortunately, people who handle customer service questions can make mistakes in some
cases. He could really forget the information he knew. ChatBot will always give you the right
answers based on the questions asked.
154
SAP CAI
Sign in at https://cai.tools.sap/ “Fig. 1”.
Fig. 1. SAP CAI Registration Screen
After registering, the bots that were previously created are displayed in the next screen. A
new chatBot project is started with the “New Bot” button. On the screen, there are templates with
ready answers against standard sentences available to users such as “Hello”, “How are you”. These
templates can also be imported into the system [2].
Fig 2. SAP CAI Register Screen
One of the language options as in Figure 3 is selected.
155
Fig 3. SAP CAI Language Options
A new bot is created with “CREATE A BOT” button.
Intents
An intent is a box of expressions that mean the same thing but are constructed in different ways.
Intents are the heart of your bot’s understanding. Each one of your intents represents an idea your
bot is able to understand “Fig. 4”.
Fig. 4. Intents
You want your bot to understand when someone asks for help. Just create an intent called
“help” and fill it with every expression a user would say when asking for guidance.”Fig. 5”:
Could you help me?
I’m lost, give me a hand please.
Can you help?
What can you do for me?
156
Fig. 5. Conversation
As in “Figure 6”, the intent, named “ask-leave”, which specifies the permission request, is
created. To use the chat screen, the message written by the staff is trained.
Fig. 6 Training Ask-Leave Intent
Entity
An entity is a keyword that is extracted from an expression. SAP CAI automatically detect 28
different entities such as Datetime, Location, Person, and so on. They are called gold entities.
157
However, there is no limit. It can also be tagged own custom entities to detect keywords depending
on bot’s context, such as subway stations if it is being built a transport assistant.
Fig. 7. Entity List
“In Fig. 8”, if the request for a permit is selected, a half-day entity is created to be asked for a
half-day or a full-time option.
Fig 8. Training #half-day entity
Condition, Api’s
A condition is a test that can be evaluated to either true or false, “Fig. 9”.
There’s a finite list of operators can be used:
Is to test equality between two values,
158
Is-not to test inequality between two values,
In to check if a value is in a list of elements,
Not-in to check if a value is not in a list of elements,
Matches to match a value with a regular expression,
Matches-not to check if the value doesn’t match with a regular expression,
Lower-than to test if the value is lower than another,
Greater-than to test if the value is greater than another,
Is-present to test if the value is present in the conversation state,
Is-asbent to test if the value is asbent in the conversation state.
Fig. 9. Usage of Condition
At many points in conversation, you most likely want to retrieve business information or
connect to an external system to perform actions. You can do this through webhooks. A webhook
is a simple HTTP call to your backend. To configure your HTTP call, click CALL WEBHOOK in
the Bot Builder “Fig. 10”.
You can provide the full URL or route (starting with a ‘/’) to be called by the Bot. Builder.
If you provide a route, the Bot webhook base URL (configurable in your bot’s settings) will be
prepended to it.
You can specify the HTTP method to use in your webhook call (GET, POST, PUT, or
PATCH).
159
Fig 10. Api
Test Bot
The generated chatbot can be tested from the bottom right window ”Fig. 11”.
Fig 11. Testing ChatBot
Building Bot
Fig 12. Bot Channels
160
The Webchat channel is developed by the SAP Conversational AI team and is an opensource project on GitHub “Fig. 12”.
You can use the default version of the webchat that we provide in the platform or customize
the open-source version by forking it and deploying it on your side.
Fig. 13. chatBot Script
Webchat channel give a script and when it is placed at html page, it can be used “Fig. 13”.
Fig. 14. ChatBot Script
Conclusion
SAP CAI product is described. The digital assistant was made. The permission request of the
personnel has been processed “Fig. 14”.
161
REFERENCES
[1] Kayla Sloan, 6 Benefits of Using Chatbots in Your Business, May 9, 2018
Https://Due.Com/Blog/Chatbots-Business-Benefits
[2] Create your chatbot. Https://Cai.Tools.Sap/Docs/Concepts/Create-Builder-Bot
[3] Michael Quoc, 11 Examples of Conversational Commerce and Chatbots, Jun 1, 2016
Https://Chatbotsmagazine.Com/11-Examples-Of-Conversational-Commerce-57bb8783d332
[4] Matt Mayer, Building a WeChat (Weixin) robot, March 22, 2014
Https://Blog.Reigndesign.Com/Blog/Building-A-Wechat-Weixin-Robot/
162
Map Ranking, Map-Reduce and Application in Big Data Analysis
Safiye TURGAY, Suat ERDOĞAN
Industrial Engineering Department, Sakarya University, TURKEY
Abstract: The map method works with a certain algorithm and the inputs which send to a value
list as a parameter. All values in the list converted to the intermediate result list. We sort all of the
data then obtain the map list. The proposed and developed map structure was tested with quicksort approach. The sorting process depends on the byte situation of the each data. Small data can
do it easily side by side. Thus, small data do not need to applying of the reduce process. Sample
selection havbe to will be easier. The goal is to give an intermediate operationtothe map-reduce
structure. The more accurate is to get a ranking. In large data analysis, the data mapping sequence
and reduction works with a certain algorithm structure, introducing and sending inputs as a
parameter to a value list. An intermediate result list is created by converting all the values in the
list, which are included in the entered system. In the structure developed after the mapping (Map)
process, the mapping list is divided and obtained. The order depends on the byte value that is
generated by each data. In case of large volume data, the data will be used without using a single
line operation. In short, the data may be side-by-side, so there is no need to apply the reduction to
each data. Therefore, the process of selecting the sample will be easier.
Keywords: Big Data, Map Reduce, Sorting, Styling.
REFERENCES
[1] Schönberger, Viktor Mayer ve Kenneth Cukier. Büyük Veri - Yaşama, Çalışma ve Düşünme
Şeklimizi Dönüştürecek Bir Devrim. Çev. Banu Erol. İstanbul: Paloma, 2013.
[2] Demchenko Y, de Laat C, Membrey P (2014) Defining Architecture Components Of the Big
Data Ecosystem. In International Conference on Collaboration Technologies and Systems (CTS) pp. 104112.
163
[3] Rabie A. Ramadan, Big Data Tools-An Overview, International Journal of Computer &
Software Engineering Volume 2 (2017), Article ID 2: IJCSE-125, 15 pages
[4] http://www.buyukveri.co/hadoop-hdfs-mimarisi/?
[5] https://hadoop.apache.org/docs/r1.2.1/mapred_tutorial.html
[6] MapReduce örnek – datascience.istanbulwww.datascience.istanbul
[7] https://dzone.com/articles/word-count-hello-word-program-inmapreduce
[8] Map ReducewithExamplesdatascienceguide.github.io
[9] http://www.dreamsyssoft.com/java-8-lambda-tutorial/map-reducetutorial.php
[10]
java
-Hadoop
map-reduceoperation
is
failing
on
writingoutput
-
Stack
Overflowstackoverflow.com
[11] Source codefororg.apache.hadoop.mapred.MapRunner.java - zGrepCodezgrepcode.com
[12] howto set configurationsfor Map-reducejobsforHadoop? - Stack Overflowstackoverflow.com
[13] Introductionto MapReducefor .NET Developers – DeveloperZendeveloperzen.com
[14] MapReduce Nedir? | B3LAB PORTALwww.b3lab.org
[15]
Introductiontothe
MapReduce
Life
Cycle
|
SUPINFO,
ÉcoleSupérieure
d'Informatiquewww.supinfo.com
[16] How to: Launching MapReduceJobs | MapRmapr.com
[17] http://salsahpc.indiana.edu/tutorial/
[18] Word Count Program With MapReduceandJava -DZoneBig Datadzone.com
[19] http://www.idearastirma.com/icerik/36/orneklemenin-onemi.aspx
[20] http://dresneradvisory.com/
[21] https://www.guru99.com/create-your-first-hadoop-program.html
[22]
ApacheHadoop-
AzureHDInsight
için
Java
MapReduce
oluşturma
|
Microsoft
Docsdocs.microsoft.com
[23] MapReduceTutorial | MapreduceExample in ApacheHadoop | Edurekawww.edureka.co
[24] https://www.kadinyazilimci.com/siralama-algoritmalari-ii/
164
Benchmarking of OECD Countries in Views of Value-Added Manufacturing Using DEA
Billur ECER, Ahmet AKTAŞ, Mehmet KABAK
Industrial Engineering Department, Gazi University, TURKEY
Abstract: Value-added is an important term which indicates the efficiency level of economic
activities. Valueadded term describes the difference between the value of produced goods and the
total cost of production. Some areas related to value-added manufacturing are pharmaceuticals,
automotive, computer and communications equipment manufacturing, aviation, etc. To develop a
value-added product some inputs are needed. Some of these inputs are energy, labor force and
research and development activities. The main aim of this study is to present a benchmarking
analysis of 29 OECD countries in views of value-added manufacturing values. To do so, an output
oriented BCC model is used to evaluate efficiency of countries and obtained results of the analysis
provide some improvement ways for countries and their position against the other countries.
Keywords: OECD Countries, Benchmarking, Value-Added Manufacturing, DEA
Introduction
Value-added term is important for manufacturing operations, because the main aim of
manufacturing is to make profit. Value-added describes the difference between the selling price of
the product and the total cost to produce it. The leading value-added manufacturing areas can be
listed as, pharmaceuticals, organic and inorganic chemical manufacturing, plastics manufacturing,
semiconductors, computer manufacturing, communications equipment manufacturing, surgical
and medical instruments manufacturing, automotive parts and aviation parts.
Almost all of the areas related with value-added manufacturing are hi-tech products. In this
regard, research and development (R&D) activities are correlated with value-added
manufacturing. There is no debate that research and development activities are conducted by
manufacturers to discover some improvement strategies for their processes to develop their
products or reduce costs.
165
Moreover, manufacturing is an activity that requires labour and energy inputs. The usage of
greater amount of energy per capita and the greater ratio of employment in industry in a country
can give us an idea for understanding the importance of manufacturing and industry in a country.
The main aim of this study is to present a benchmarking analysis of value-added
manufacturing values of 29 OECD countries by considering R&D activities, industrial
employment rates and energy usage. To perform this analysis Data Envelopment Analysis (DEA)
technique is used. Output-oriented BCC (Banker, Charnes, Cooper) model is chosen for the
analysis and obtained results of the model are discussed.
The rest of the paper organized as follows: a literature review for recent studies on
benchmarking of countries by using DEA models is presented in the second part. Next, basic
definitions for DEA and BCC model are given in the third part. The fourth part contains the
information and the results about the benchmarking application of countries. Finally, the paper is
concluded in the fifth part by presenting conclusions and future research directions.
Recent Literature of Benchmarking of Countries
Benchmarking of countries is a popular research area among researchers. In this part, a summary
of 12 studies published between 2016 and 2019 is presented. DEA models are utilized in all of
these studies. Details of these studies are presented as follows:
Chen and Hung analyzed efficiency of R&D activities in 25 countries by using network DEA
model [1]. They develop a three-stage efficiency model consisting research, translation and
economic efficiency stages.
Timmer et al. [2] developed a decomposition framework based on DEA to analyze labor
efficiency difference between Germany and USA in the early 20th century. Results of their
analyses lead them to obtain some findings to understand the drivers behind the productivity
differences.
Environmental performance efficiency of European countries is evaluated in Chodakowska
and Nazarko’s study [3]. By using DEA models, environmental performance and technological
competition indicators are integrated. They found out the diversification of environmental
performance of European countries.
166
Efficiency of CO2 emission reduction techniques applied on 12 European countries are
compared by a two-stage DEA model [4]. Kwon et al.’s analysis shows benchmarks for inefficient
countries to improve their CO2 reduction strategies.
Marti et al. [5] developed a DEA based logistics performance index (LPI) approach to
benchmark countries in terms of logistics performance. They analyzed a number of scenarios by
considering different combinations of factors that affecting LPI.
Storto and Goncharuk’s study presents a benchmarking analysis of European countries in
views of efficiency and effectiveness of healthcare services [6]. They used slack based measure
model of DEA to identify shortcomings of national healthcare systems.
Faghih et al. [7] used DEA models to determine benchmarks of 55 MENA countries in views
of national entrepreneurial efficiency. The data used in the analysis is gathered from Global
Entrepreneurship Monitor data.
Road safety policies across Europe is evaluated by Nikolaou and Dimitriou [8]. They used
data of 2005 – 2014 for 23 EU countries and discover some managerial implications from the
analysis via DEA.
Output oriented BCC model is used in See and Yen’s study to analyze efficiency of health
services [9]. Data of 121 countries are evaluated in the analysis and benchmarks are presented for
policy makers.
Network DEA model is used in Wanke et al.’s study to investigate drivers of railway
performance in selected Asian countries [10]. Different improvement suggestions for countries are
provided as a result of the analysis.
Efficiency of government excellence in 45 low and low-middle income countries are
evaluated by Choi and Park [11]. Effect of government excellence on social progress is
investigated in the study and countries with same economic conditions are benchmarked with each
other.
Logistic performance evaluation of OECD countries is considered in Rashidi and Cullinane’s
study [12]. Top performers and inefficient countries are presented with benchmarking countries at
the end of the analysis.
167
It is seen that different DEA models are used to compare efficiency of countries. Therefore,
DEA is an appropriate method for our analysis.
Methodology
Efficiency of a process is defined as the ratio of outputs to the inputs. It describes the degree of
transformation of inputs to the outputs. The higher value that efficiency takes, the more efficient
the process is. Efficiency of process with more than one input and/or output is calculated by
considering importance degree of the inputs and outputs. DEA is a technique, which is used for
measuring the relative efficiency of a number of systems with the same inputs and outputs. DEA
models try to determine the weights of inputs and outputs for determining the best efficiency score
of each decision-making unit (DMU).
There are two different problem assumptions for measuring efficiency with DEA models. The
system that is under consideration can aim the same output level with lower input (input-oriented)
or obtain more output with the same input level (output-oriented).
Moreover, systems may follow constant return to scale (CRS) or variable return to scale (VRS)
assumptions. CRS is about existence of a constant efficiency frontier, while VRS accepts
efficiency frontier may change related to the input level. Based on the scale assumption, there are
different models named as CCR (Charnes, Cooper, Rhodes) and BCC (Banker, Charnes, Cooper).
In this study, value added manufacturing efficiency of countries is measured by using DEA
model. It is thought that VRS assumption is appropriate for this problem, because of the different
development levels of countries. Also, a DMU with higher output is more efficient than a DMU
which has the same level of input. So, the DEA model for this study is determined as outputoriented BCC model. The mathematical formulation of this model is given as follows [13]:
max zk
n
Yrj jk Yrk zk 0
r
j 1
n
x
j 1
ij
n
j 1
jk
jk
X ik
i
1
jk 0
j
168
Benchmarking Analysis of OECD Countries
In this study, value added manufacturing efficiency of 29 OECD member countries are measured
by using a DEA model. BCC model is used to measure efficiency of countries. The output of this
efficiency analysis is determined as percent ratio of value-added manufacturing in the country’s
GDP (MVA). On the other hand, percentage of employment in industry of total employment (EiI),
energy use per capita (kg of oil equivalent) (EU) and research and development expenditure
percentage of GDP (RD) are considered as inputs of value-added manufacturing. Data of 29 OECD
member countries are retrieved from World Bank’s database of World Development Indicators
[14] for 2015. These data are the most recent data of countries considered in the analysis. The other
member countries’ data are not available, that is why they are out of consideration. Countries that
taken into consideration in this study are Austria, Belgium, Chile, Czech Republic, Denmark,
Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Israel, Italy, Japan, Korea Republic,
Luxembourg, Mexico, Netherlands, Norway, Poland, Portugal, Slovak Republic, Slovenia, Spain,
Sweden, Turkey, United States and United Kingdom. Data of the efficiency analysis are given in
Table 1 as follows:
TABLE 1. DATA FOR THE ANALYSIS
Country
Indicator
MVA
EiI
EU
RD
AUT
16.56
25.76
3800.30
3.05
BEL
12.83
21.43
4687.79
2.47
CHL
11.56
23.28
2028.89
0.38
CZE
24.13
38.03
3860.00
1.93
DNK
12.41
19.29
2816.61
2.96
EST
13.83
30.66
4173.33
1.49
FIN
14.80
21.70
5924.70
2.90
FRA
10.43
20.38
3689.52
2.27
DEU
20.76
27.69
3817.55
2.92
GRC
8.31
14.94
2182.07
0.97
HUN
20.52
30.33
2432.75
1.36
169
Country
Indicator
MVA
EiI
EU
RD
IRL
34.32
19.08
2819.88
1.20
ISR
11.78
17.71
2777.88
4.27
ITA
14.39
26.60
2481.75
1.34
JPN
20.72
25.91
3428.56
3.29
KOR
27.09
25.07
5413.35
4.22
LUX
4.86
12.52
6548.41
1.27
MEX
17.14
25.06
1488.02
0.53
NLD
10.70
16.43
4233.04
2.00
NOR
6.86
20.12
5815.81
1.93
POL
17.64
30.54
2490.21
1.00
PRT
12.16
24.35
2131.68
1.24
SVK
19.75
36.11
3003.66
1.18
SVN
19.96
32.01
3174.87
2.20
ESP
12.90
19.90
2571.34
1.22
SWE
13.69
18.29
5102.79
3.27
TUR
16.71
27.23
1656.80
0.88
USA
11.92
18.85
6797.62
2.74
GBR
8.97
18.66
2763.98
1.67
DEAP software is used measure efficiency of countries. Efficiency analysis results of
output-oriented BCC model is given in Table 2.
According to efficiency analysis results in Table 2, Ireland and Mexico are the efficient
countries under constant return to scale assumption among 29 countries. Efficient countries for
variable return to scale assumption are determined as Chile, Greece, Ireland, Luxemburg and
Mexico. As it mentioned before, under VRS assumption, efficiency frontier changes for different
input levels and number of efficient DMUs are generally greater than number of efficient DMUs
in CCR models. Scale efficiency is the ratio of CCR efficiency to BCC efficiency and it shows the
DMU has increasing return to scale or decreasing return to scale.
170
TABLE 2. EFFICIENCY VALUES
Country
EFFICIENCY SCORE
CCR
BCC
SCALE
AUT
0.358
0.483
0.742 (DRS)
BEL
0.333
0.374
0.890 (DRS)
CHL
0.941
1.000
0.941 (IRS)
CZE
0.514
0.703
0.731 (DRS)
DNK
0.362
0.362
1.000 -
EST
0.318
0.403
0.790 (DRS)
FIN
0.379
0.431
0.879 (DRS)
FRA
0.285
0.304
0.936 (DRS)
DEU
0.447
0.605
0.739 (DRS)
GRC
0.313
1.000
0.313 (IRS)
HUN
0.693
0.700
0.990 (IRS)
IRL
1.000
1.000
1.000 -
ISR
0.370
0.451
0.820 (IRS)
ITA
0.476
0.480
0.992 (IRS)
JPN
0.497
0.604
0.822 (DRS)
KOR
0.601
0.789
0.761 (DRS)
LUX
0.216
1.000
0.216 (IRS)
MEX
1.000
1.000
1.000 -
NLD
0.362
0.482
0.751 (IRS)
NOR
0.190
0.200
0.948 (DRS)
POL
0.593
0.604
0.981 (DRS)
PRT
0.469
0.478
0.981 (IRS)
SVK
0.554
0.584
0.949 (DRS)
SVN
0.517
0.582
0.888 (DRS)
ESP
0.412
0.425
0.970 (IRS)
SWE
0.416
0.445
0.935 (IRS)
TUR
0.829
0.865
0.958 (IRS)
USA
0.352
0.358
0.982 (IRS)
171
EFFICIENCY SCORE
Country
CCR
BCC
SCALE
GBR
0.267
0.283
0.945 (IRS)
MEAN
0.485
0.586
0.857
Another important result obtained with DEA is the reference sets and target values for
inefficient DMUs. These values are presented in Table 3. For example, Turkey should be similar
to Ireland and Mexico with weight values of 0.127 and 0.873. That means Turkey has to change
their input and output values to sum of 0.127 of Ireland’s scores and 0.873 of Mexico’s scores.
That change will bring Turkey’s scores into an output value of 19.317 and input values of 24.302,
1656.800 and 0.615, respectively.
TABLE 3. REFERENCE SETS AND TARGET VALUES
Country
Indicator
REFERENCE SET
MVA
EiI
EU
RD
AUT
IRL (1.000)
34.320
19.080
2819.880
1.200
BEL
IRL (1.000)
34.320
19.080
2819.880
1.200
CHL
CHL (1.000)
11.560
23.280
2028.890
0.380
CZE
IRL (1.000)
34.320
19.080
2819.880
1.200
DNK
IRL (0.998), MEX (0.002)
34.278
19.095
2816.610
1.198
EST
IRL (1.000)
34.320
19.080
2819.880
1.200
FIN
IRL (1.000)
34.320
19.080
2819.880
1.200
FRA
IRL (1.000)
34.320
19.080
2819.880
1.200
DEU
IRL (1.000)
34.320
19.080
2819.880
1.200
GRC
GRC (1.000)
8.310
14.940
2182.070
0.970
HUN
IRL (0.709), MEX (0.291)
29.326
20.818
2432.750
1.005
IRL
IRL (1.000)
34.320
19.080
2819.880
1.200
ISR
GRE (0.274), IRL (0.690), LUX (0.036)
26.132
17.710
2777.880
1.139
ITA
IRL (0.746), MEX (0.254)
29.958
20.598
2481.750
1.030
JPN
IRL (1.000)
34.320
19.080
2819.880
1.200
172
Country
Indicator
REFERENCE SET
MVA
EiI
EU
RD
KOR
IRL (1.000)
34.320
19.080
2819.880
1.200
LUX
LUX (1.000)
4.860
12.520
6548.410
1.270
MEX
IRL (1.000)
17.140
25.060
1488.020
0.530
NLD
GRE (0.031), IRL (0.585), LUX (0.384)
22.189
16.430
4233.040
1.220
NOR
IRL (1.000)
34.320
19.080
2819.880
1.200
POL
IRL (0.701), MEX (0.299)
29.192
20.865
2422.310
1.000
PRT
IRL (0.483), MEX (0.517)
25.443
22.170
2131.680
0.854
SVK
IRL (0.970), MEX (0.030)
33.807
19.259
2780.123
1.180
SVN
IRL (1.000)
34.320
19.080
2819.880
1.200
ESP
GRE (0.042), IRL (0.791), MEX (0.166)
30.363
19.900
2571.340
1.079
SWE
IRL (0.880), LUX (0.120)
30.772
18.290
3268.895
1.208
TUR
IRL (0.127), MEX (0.873)
19.317
24.302
1656.800
0.615
USA
IRL (0.965), LUX (0.035)
33.287
18.850
2950.606
1.202
GBR
GRE (0.099), IRL (0.900), LUX (0.002)
31.703
18.660
2763.980
1.177
Conclusion
Value added manufacturing percentage of GDP is one of the important indicators for
competitiveness in global trade competition between countries. In this study, efficiency of valueadded manufacturing values of 29 OECD member countries are measured by using output-oriented
BCC model. Mean efficiency of 29 countries is measured as 0.586 and efficient countries are found
as Chile, Greece, Ireland, Luxemburg and Mexico. Benchmarks for inefficient countries are
presented and target values are calculated by DEAP software.
In further studies, this study can be extended by considering different indicators in the
model. Moreover, Malmquist total factor productivity index can be calculated by considering
different years’ data to analyze changes of efficiency in a multi-period view.
REFERENCES
[1] P. C. Chen And S. W. Hung, “An Actor-Network Perspective on Evaluating The R&D Linking
Efficiency Of İnnovation Ecosystems,” Technological Forecasting & Social Change, 112, 303–312, 2016.
173
[2] M. P. Timmer, J. Veenstra And P. J. Woltjer, “The Yankees of Europe? A New View On
Technology and Productivity in German Manufacturing İn The Early Twentieth Century,” The Journal Of
Economic History, 76, 874-908, 2016.
[3] E. Chodakowska and J. Nazarko, “Environmental Dea Method For Assessing Productivity Of
European Countries,” Technological And Economic Development of Economy, 23, 589–607, 2017.
[4] D. S. Kwon, J. H. Cho And S. Y. Sohn, “Comparison of Technology Efficiency For Co2
Emissions Reduction Among European Countries Based on Dea With Decomposed Factors,” Journal Of
Cleaner Production, 151, 109-120, 2017.
[5] L. Martí, J. C. Martín And R. Puertas, “A Dea-Logistics Performance Index,” Journal of Applied
Economics, 20, 169-192, 2017.
[6] C. L. Storto And A. G. Goncharuk, “Efficiency Vs Effectiveness: A Benchmarking Study on
European Healthcare Systems. Economics And Sociology, 10, 102-115, 2017.
[7] N. Faghih, M. R. Zali And N. Vafaei, “Entrepreneurial National Efficiency Based on Gem Data:
Benchmarks For The Mena Countries” In: Faghih N., Zali M. (Eds) Entrepreneurship Ecosystem in The
Middle East And North Africa (Mena). Contributions To Management Science. Springer, Cham.
[8] P. Nikolaou And L. Dimitriou, “Evaluation of Road Safety Policies Performance Across
Europe: Results From Benchmark Analysis For A Decade,” Transportation Research Part A, 116, 232–246,
2018.
[9] K. F. See And S. H. Yen, “Does Happiness Matter To Health System Efficiency? A Performance
Analysis,” Health Economics Review, 8, 33, 2018.
[10] P. Wanke, Z. Chen, W. Liu, J. J. M. Antunes And M. A. K. Azad, “Investigating The Drivers
Of Railway Performance: Evidence From Selected Asian Countries,” Habitat International, 80, 49–69,
2018.
[11] H. Choi And M. J. Park, “Evaluating The Efficiency Of Governmental Excellence For Social
Progress: Focusing on Low And Lower-Middle-İncome Countries,” Social Indicators Research, 141, 111–
130, 2019.
[12] K. Rashidi And K. Cullinane, “Evaluating The Sustainability of National Logistics
Performance Using Data Envelopment Analysis,” Transport Policy, 74, 35–46, 2019.
[13] A. Tarım, “Veri Zarflama Analizi: Matematiksel Tabanlı Göreli Etkinlik Ölçüm Yaklaşımı,”
Ankara: T.C. Sayıştay Başkanlığı, Araştırma Çeviri İnceleme Dizisi, 2001.
[14] Internet, “Http://Data.Worldbank.Org/Data-Catalog/World-Developmentindicators”, Access
Date: 20.04.2019
174
A Study on Method Prediction for a Better Directed Treatment of Warts
Ahmet ÇİFCİ1, Mehmet ŞİMŞİR2
1
Deparment of Electrical-Electronics Engineering, Burdur Mehmet Akif Ersoy University,
2
Department of Mechatronics Engineering, Karabük University, TURKEY
Abstract: Various kinds of medical treatment methods may be used to cure common types of
diseases. Experience based predictions are done to choose a treatment method among the choices
of cures to get better results for the patient. This condition sometimes may continue with trying
another treatment method unless a satisfactory result is reached and changing the treatment method
of the cure process is not a desired course for time and health. This study presents a confident way
to choose the treatment method for wart disease by using feedforward neural network. The study
uses two types of datasets, one for cryotherapy and other for immunotherapy treatment methods.
It was observed from the experimental results that, feedforward neural network achieved 94.4%
success and 85.6% success for cryotherapy and immunotherapy datasets, respectively. The results
are remarkable for both doctors and patients.
Keywords: Cryotherapy, Immunotherapy, Feedforward Neural Network, Wart.
Introduction
Warts are benign skin growth caused by the virus called Human papillomavirus (HPV), which
infect the top layer of the skin [1]. Most warts are harmless, but they are highly contagious.
Traumas events such as cuts or damages to the skin facilitate infection. They are spread so quickly,
especially in the summer. Direct contact with a wart or contacts with another person by sharing
towels, razors, or other personal items may cause the virus to spread [2]. Although they are most
common on the knuckles, fingers, hands, elbows, knees, they may occur in the whole body.
Children, adolescents, people who bite their nails, and people with a weak immune system have a
higher risk of developing warts [3]. Warts are usually flesh-coloured, hard and rough. On the other
hand, there are dark (brown, grey-black), flat and soft wart types.
175
Types of wart include common warts (verruca vulgaris), plantar warts (verruca plantaris),
flat warts (verruca plana), and genital warts (condylomata acuminata) [4]. Common warts are
most commonly seen in the hands, between the fingers, legs, and around the nails [5]. Plantar warts
are embedded in the skin and can be painful. Plantar warts can be confused with calluses [6]. Even
though flat warts can occur on any part of the body, they are often found on the face and hands.
They are small, smooth and flesh-coloured. They can occur in large numbers [4]. Genital warts are
soft, flesh‐coloured papules on the genitalia and breech. It is more common in people who have
sexual activity without a condom [4].
There exist several methods available for the treatment of warts. Salicylic acid treatment,
electrosurgery, freezing (cryotherapy), immunotherapy, and laser treatment are the main methods
to get rid of warts [7]. The most common methods for the treatment of warts are cryotherapy and
immunotherapy. Cryotherapy is performed by applying liquid nitrogen gas to the desired skin
lesion using special devices. Liquid nitrogen gas at a temperature of -196 degree Celsius will freeze
the tissue. This freezing process is a short duration of time (10 - 60 seconds); at the end of this
time, the tissue will return to its normal temperature. In this short-term and immediate freezingmelting process, the cells in the targeted tissue will be destroyed and will die. These dead and
abnormal cells, which no longer function, will be removed from the tissue during the healing
process and replaced by fresh tissues [8,9]. Immunotherapy is a new type of treatment that aims to
prevent or eliminate the growth of cancer cells that abnormally proliferate (such as HPV lesions).
Immunotherapy uses the patient's own immune system to fight warts. The antigen is injected into
the body to activate the immune system [10,11]. Cryotherapy is an uncomplicated method for the
patients with respect to immunotherapy.
Numerous investigations have been reported in the literature on the treatment of warts. In
[12], Putra et al. used AdaBoost and Random Forest as a strong learner or a weak learner for
selection of wart treatment method. The results showed that the accuracy rate of cryotherapy was
96.6% and accuracy rate of immunotherapy was 91.1%. Tanyıldızı et al. [13] judged the
performances of classification algorithms in cryotherapy and immunotherapy datasets. The best
success rate obtained using the K-star algorithm is found as 96.66% for cryotherapy and using the
Random Forest algorithm is found as 85.55% for immunotherapy. Fuzzy Rule, Naive Bayes, and
Random Forest based algorithms have been carried out in cryotherapy and immunotherapy
treatment for comparing the effectiveness of these algorithms by Akyol et al. in [14]. They
176
concluded that the random forest algorithm outperforms other classification algorithms in both
accuracy and sensitivity within cryotherapy and immunotherapy datasets. Nugroho et al. proposed
C4.5 algorithm combined with Random Forest Feature Weighting for wart treatment selection
method [15]. The results showed that the proposed method can improve the performance of
prediction. Ali et al. applied some algorithms to show that which treatment is more effective from
cryotherapy and immunotherapy [16]. They concluded that cryotherapy treatment is better than
immunotherapy. See also [17], where the author used the decision tree-based method to specify
the rules of predicting the performance of wart treatment methods. The results obtained the level
of accuracy of 94.4% on cryotherapy and 90% on immunotherapy.
In this study, feed-forward neural network was used to decide and select the cure for wart
treatment. Patient data were used as input and success of cure type was used as output data. A
neural network for immunotherapy and another one for cryotherapy was built to decide which
methodology is appropriate for the patient. As it was mentioned before, cryotherapy is an
uncomplicated method for the patient with respect to immunotherapy. When a patient data is
applied to neural network for cryotherapy and gives positive outputs for applicability. Treatment
method starts with cryotherapy cure, but if cryotherapy neural network gives ineffective result as
an output, applying data to immunotherapy neural network is preferred and this situation supports
a logical way to select the uncomplicated method at first without waste of time. Because
immunotherapy method has a higher result of success percentage according to the datasets used in
this study. But this method may also be abrasive for the patient.
The paper is structured in the following manner. Section 2 presents the immunotherapy and
the cryotherapy datasets. Method and experimental results are given in section 3. Finally,
conclusions are duly drawn in section 4.
Dataset
The immunotherapy and the cryotherapy datasets used in this study are gathered from the
University of California, Irvine (UCI) Machine Learning Repository [18,19]. The datasets are
collected along two years from the dermatology clinic of Ghaem Hospital in Mashhad, Iran
[20,21]. The immunotherapy with candida antigen and the cryotherapy with liquid nitrogen were
applied to 180 patients with plantar and common warts. Each data set contains 90 patients. Patients
were randomly selected. The datasets do not have any missing value.
177
The immunotherapy dataset consists of eight features; gender, age, time elapsed before
treatment, the number of warts, types of wart, surface area of warts, induration diameter of initial
test, and response to treatment. The details of the immunotherapy dataset are presented in Table 1.
Table 1. Features of Immunotherapy Dataset
Feature No.
Feature Name
Values
1
Gender
2
Age (year)
15-56
3
Time elapsed before treatment (month)
0-12
4
The number of warts
1-19
41 Male
49 Female
47 Common
5
Types of wart (count)
22 Plantar
21 Both
6
Surface area of warts (mm2)
6-900
7
Induration diameter of initial test (mm)
5-70
8
Response to treatment
Yes or No
The cryotherapy dataset has seven features; gender, age, time elapsed before treatment,
the number of warts, types of wart, surface area of warts, and response to treatment. The details
of the cryotherapy dataset can be seen in Table 2.
Table 2. Features of Cryotherapy Dataset
Feature No.
Feature Name
Values
1
Gender
2
Age (year)
3
Time elapsed before treatment (month) 0-12
4
The number of warts
47 Male
43 Female
15-67
178
1-12
54 Common
5
Types of wart (count)
9 Plantar
27 Both
6
Surface area of warts (mm2)
4-750
7
Response to treatment
Yes or No
Method and Experimental Results
With the developments on intelligent systems, human expertise can be simulated by artificial
neural networks to gain time and to get more accurate decisions. As a general approach, our study
is based on previously taken and reliable reference data. After a general overview on types of cures
for wart treatment, it is necessary to classify the patient data. “Gender”, “age”, “time elapsed before
treatment”, “the number of warts”, “types of wart” and “surface area of warts” features for the
patients were used as the classification titles for the neural network data inputs. And output data
was also used as success of cure type as “0” and “1”.
A feedforward neural network was chosen for this study. Feedforward neural networks
has input, output and at least one hidden layers. The neural network was constructed with one
hidden layer. If too many hidden layers and neurons are used, overfitting may occur, that is
although the network can be trained to work very well for the training data, it performs poorly
for test data. It is essential to optimize the success percentage of the neural network with
effectual numbers of neurons for the hidden layer to achieve the best performance level [22,23].
The constructed neural networks have 6 input variables, 6 neurons for the hidden layer and
also 2 for output for both cryotherapy and immunotherapy as seen in Fig. 1.
179
Fig. 1. General Structure Of Constructed Feed Forward Backpropagation Neural Networks For Both
Cryotherapy And Immunotherapy.
To determine the best condition for a better success percentage of neural networks, several
numbers of neuron numbers were experimented by MATLAB and current configuration shown in
Fig. 1 was obtained because of better success rate according to the tested other choices. The
training performance curves of the designed neural networks are shown in Fig. 2 for cryotherapy
and Fig. 3 for immunotherapy. Best validation level of the neural networks for training process is
indicated by means of training, validation and test curves. The training process of the neural
network was reached to goal at 7 epochs in terms of minimum gradient for cryotherapy neural
network.
180
Fig. 2. Training Performance Curves of Neural Network Designed For Cryotherapy
by MATLAB.
Fig. 3. Training Performance Curves of Neural Network Designed For Immunotherapy
By MATLAB.
The training state throughputs of the neural network are shown in Fig. 4 for cryotherapy
and Fig. 5 for immunotherapy, which indicate weight changes and validation checks for the neural
networks and gradients till the determined goals were reached.
181
F g. 4. Training State Data of Neural Network For Cryotherapy by MATLAB.
F g. 5. Training State Data of Neural Network For Immunotherapy by MATLAB.
After acceptable satisfactory levels were reached according to the number of data used, it
was necessary to test the success rate of the designed and trained neural networks by the help of
MATLAB Simulink model as seen in Fig. 6. The system in Fig. 6 was designed for testing the
constructed neural networks using the pre-paired patient data to achieve a performance test.
182
F g. 6. MATLAB Simulink Model For Neural Network Performance Test.
After performance test for cryotherapy neural network, a success percentage of 94.44%
was achieved. And the neural network test results for immunotherapy was 85.6% after the same
performance test which was realized using the model seen in Fig. 6. Both cryotherapy and
immunotherapy network trainings were practiced using 90 sets of patient data but there is a success
percentage difference between them. The cause of this difference is the success level of
immunotherapy method. Immunotherapy is more successful with respect to cryotherapy according
to the datasets used in this study, it is also a more onerous cure for wart treatment. And this rate of
success for cure treatment decreases the variety of outputs as a result reduction. As a result of this
situation, the success percentage of the neural network was directly affected.
As it was mentioned before, to find a suitable cure for wart treatment is important. And it
is also a noteworthy point that; trying the easier and more effortless cure for the patient is a big
advantage. Although the onerous method has a lower level of neural network success percentage.
There is a confident way to pre-test the cure type with the neural network to prevent loss of time.
This situation will also increase the success percentage and usage numbers of the more effortless
cure types for treatments by pre-testing them firstly instead of directly using onerous cures of
treatments.
Conclusion
In this study, the most common wart treatment methods, cryotherapy and immunotherapy, were
analysed for wart treatment prediction by applying a feedforward neural network. The
experimental results show that the successes of feedforward neural network were 94.4% and 85.6%
183
for cryotherapy and immunotherapy methods respectively. Although acceptable levels of success
percentages were obtained by the designed neural networks, it is also possible to achieve better
levels of success percentages for the neural networks using more patient data while training the
designed neural networks.
This study takes an inspiring role in the cure selecting for wart treatment by obtaining
positive results for a better directed preference. Using artificial intelligence for cure selections of
treatments will cause obtaining faster and well directed treatment processes.
REFERENCES
[1] K. E. Rowson, and B. W. Mahy, “Human papova (wart) virus,” Bacteriological Reviews, vol.
31, no. 2, pp. 110–131, 1967.
[2] D. C. Wiley, and A. C. Cory, Encyclopedia of School Health, Thousand Oaks, CA: SAGE
Publications, 2013.
[3] S. K. Loo, and W. Y. Tang, “Warts (non-genital),” BMJ Clinical Evidence, 1710, 2009.
[4] L. Allison, and M. D. Holm, “Warts,” in Pediatric Clinical Advisor, vol. II, L. C. Garfunkel, J.
Kaczorowski, and C. Christy, Eds. St. Louis: Mosby Elsevier, 2007.
[5] W. Bonnez, “Papillomaviruses,” in Mandell, Douglas, and Bennett’s Principles and Practice of
Infectious Diseases, vol. VIII; J. E. Bennett, R. Dolin, and M. J. Blaser, Eds. Philadelphia: Saunders, an
Imprint of Elsevier Inc., 2015.
[6] S. M. Simons, And R. Kennedy, “Foot Injuries,” in Clinical Sports Medicine Medical
Management And Rehabilitation, W. R. Frontera, S. A. Herring, L. J. Micheli, And J. K. Silver, Eds.
Philadelphia: Saunders Elsevier, 2007.
[7] N. H., Boroujeni, And F. Handjani, “Cryotherapy Versus Co2 Laser in The Treatment of Plantar
Warts: A Randomized Controlled Trial,” Dermatology Practical & Conceptual, Vol. 8, No. 3, Pp. 168-173,
2018.
[8] S. Cockayne, C. Hewitt, K. Hicks, S. Jayakody, A.R. Kang’ombe, E. Stamuli, Et Al.
“Cryotherapy Versus Salicylic Acid For The Treatment Of Plantar Warts (Verrucae): A Randomised
Controlled Trial,” Bmj, 342, P. D3271, 2011.
184
[9] M. Barrett, N. Harter, And A. Wysong, “Common Pediatric Dermatologic Surgery Procedures,”
in Pediatric Dermatologic Surgery, K. Nouri, L. Benjamin, J. Alshaiji, And J. Izakovic, Eds. Hoboken:
Wiley & Blackwell, 2019, Pp. 73–92.
[10] Y. C. T. Shih, L. S. Elting, A. L. Pavluck, A. Stewart, And M. T. Halpern, “Immunotherapy
İn The İnitial Treatment Of Newly Diagnosed Cancer Patients: Utilization Trend And Cost Projections For
Nonhodgkin's Lymphoma, Metastatic Breast Cancer, And Metastatic Colorectal Cancer,” Cancer Invest,
Vol. 28, No. 1, Pp. 46–53, 2010.
[11] S. Garg, And S. Baveja, “Intralesional İmmunotherapy For Difficult to Treat Warts With
Mycobacterium W Vaccine,” Journal Of Cutaneous And Aesthetic Surgery, Vol. 7, No. 4, Pp. 203–208,
2014.
[12] M. A. Putra, N. A. Setiawan, And S. Wibirama, “Wart Treatment Method Selection Using
Adaboost With Random Forests As A Weak Learner,” Communications in Science And Technology, Vol.
3, No. 2, Pp. 52–56, 2018.
[13] E. Tanyıldızı, M. Karabatak, G. Yıldırım, And Z. Özpolat, “Siğil Tedavisinde Sınıflandırma
Algoritmalarının Performans Analizi,” Fırat Üniversitesi Mühendislik Bilimleri Dergisi, Vol. 30, No. 2,
Pp. 249– 256, 2018.
[14] K. Akyol, A. Karacı, And Y. A. Gültepe, “Study On Prediction Success of Machine Learning
Algorithms For Wart Treatment,” International Conference on Advanced Technologies, Computer
Engineering And Science, May 11-13, 2018 Safranbolu, Turkey, Pp. 186–188.
[15] H. W. Nugroho, T. B. Adji, And N. A. Setiawan, “Random Forest Weighting Based Feature
Selection For C4.5 Algorithm on Wart Treatment Selection Method,” International Journal On Advanced
Science, Engineering And Information Technology, Vol. 8, No. 5, Pp. 1858–1863, 2018.
[16] Y. Ali, M. Shahzad, And A. Akbar, “Comparison Of Cryotherapy and Immunotherapy in
Warts Treatment,” International Journal Of Multidisciplinary Sciences And Engineering, Vol. 9, No. 7, Pp.
14–18, 2018.
[17] S. B. Akben, “Predicting The Success Of Wart Treatment Methods Using Decision Tree Based
Fuzzy İnformative Images,” Biocybern. Biomed. Eng., Vol. 38, Pp. 819–827, 2018.
[18] Https://Archive.İcs.Uci.Edu/Ml/Datasets/Immunotherapy+Dataset
[19] Https://Archive.İcs.Uci.Edu/Ml/Datasets/Cryotherapy+Dataset+
185
[20] F. Khozeimeh, R. Alizadehsani, M. Roshanzamir, A. Khosravi, P. Layegh, And S. Nahavandi.
“An Expert System For Selecting Wart Treatment Method,” Computers in Biology And Medicine, Vol. 81,
Pp. 167–75, 2017.
[21] F. Khozeimeh, F. Jabbari Azad, Y. Mahboubi Oskouei, M. Jafari, S. Tehranian, R.
Alizadehsani, and P. Layegh, “Intralesional İmmunotherapy Compared To Cryotherapy in The Treatment
of Warts,” International Journal Of Dermatology, Vol. 56, Pp. 474–478, 2017.
[22] L. Wang, H. C. Quek, Hou, K. H. Tee, N. Zhou, And C. Wan, “Optimal Size Of A Feedforward
Neural Network: How Much Does it Matter?,” Joint International Conference on Autonomic And
Autonomous Systems And International Conference On Networking And Services Icas/Icns, October 2328, 2005 Papeete, Tahiti, French Polynesia, P. 69.
[23] M. Şimşir, R. Bayır, And Y. Uyaroğlu, “Real-Time Monitoring and Fault Diagnosis of A Low
Power Hub Motor Using Feedforward Neural Network,” Computational Intelligence and Neuroscience,
Vol. 2016, 2016.
186
Estimation of the Demand for the Blood Bank Using Hybrid PCA-ANFIS Method
Seda Hatice GÖKLER, Semra BORAN
Department of Industrial Engineering, Sakarya University, TURKEY
Abstract: Blood is a vital product that is needed by thousands of people every day due to diseases,
surgeries or injuries. Blood banks should accurately determine the amount of blood they should
have in their stock to meet blood needs. Therefore, having less blood than necessary in hospitals
creates important problems such as not meet need for blood and loss of life. On the other hand,
storing large amounts of blood causes deteriorating the blood and causes stock out in other
hospitals. The aim of this study is to determine the criteria affecting blood demand and to forecast
the blood demand by the machine learning algorithm Adaptive Network Based Fuzzy Inference
System (ANFIS) method. However, since the number of impact criteria is high, principal
component analysis (PCA) method has been used in order to decrease criteria and eliminate the
dependencies between the criteria. The developed hybrid method was applied in a regional blood
center.
Keywords: ANFIS, PCA, Demand Forecasting, Blood Banking
REFERENCES
[1] Özdamar K. Paket Programlar ile İstatistiksel Veri Analizi-2 (Çok Değişkenli Analizler),
5.Baskı, Eskişehir, Kaan Kitabevi, 2004.
[2] Jang J. ANFIS: Adaptive-Network-Based Fuzzy İnference System. IEEE. 23(3), 665-685, 1993.
[3] Tabachnick BG, Fidell LS. Using Multivariate Statistics. Boston: Allyn and Bacon. 2001.
[4] Çokluk Ö, Şekercioğlu G, Büyüköztürk Ş. Sosyal Bilimler için Çok Değişkenli İstatistik: SPSS
ve Lisrel Uygulamaları, Pegem Akademi Yayıncılık, 2012.
[5] Ismail FH, Aziz MA, Hassanien AE. Optimizing The Parameters of Sugeno Based Adaptive
Neuro Fuzzy Using Artificial Bee Colony: A Case Study on Predicting The Wind Speed. Proceedings Of
The Federated Conference On Computer Science And Information Systems. 2015; 8: 645–651.
187
[6] Moosavi V, Vafakhah M, Shirmohammadi B, Ranjbar M. Optimization of Wavelet-ANFIS and
Wavelet-ANN Hybrid Models by Taguchi Method for Groundwater Level Forecasting Arabian Journal For
Science And Engineering. 2014; 39(3): 1785–1796.
[7] Fragiadakis NG, Tsoukalas VD, Papazoglou VJ. An Adaptive Neurofuzzy Inference System
(ANFIS) Model For Assessing Occupational Risk İn The Shipbuilding İndustry. Safety Science 2014; 63:
226–235.
[8] Alizadeh M, Lewis M, Zarandi MHF, Jolai F. Determining Significant Parameters in the Design
of ANFIS. 2011 Annual Meeting of the North American Fuzzy Information Processing Society 18-20
March 2011 El Paso, TX, USA
[9] Qu N, Zhu M, Ren Y, Dou S. Adaptive Neuron-Fuzzy İnference System Combined with
Principal Components Analysis For Determination Of Compound Thiamphenicol Powder On Nearİnfrared Spectroscopy, Journal of the Taiwan Institute of Chemical Engineers 2012; 43:566–572.
[10] Rençber ÖF. Sınıflandırma Problemlerinde Çoklu Lojistik Regresyon, Yapay Sinir Ağı ve
Anfis Yöntemlerinin Karşılaştırılması, Seçkin Yayıncılık 2018.
[11] Phootrakornchai W., Jiriwibhakorn S. Online Critical Clearing Time Estimation Using An
Adaptive Neuro-Fuzzy İnference System (ANFIS), International Journal of Electrical Power & Energy
Systems 2015, 73: 170-181.
188
Analysing of Multivariate Processes with Machine Learning Algorithms
Deniz DEMİRCİOĞLU DİREN, Semra BORAN, Seda Hatice GÖKLER
Department of Industrial Engineering, Sakarya University, TURKEY
Abstract: It is often not easy to obtain results from complex processes multi variables. Additional
techniques and methods are needed to guide. In this study, after the detecting the out of control
and under control samples with Hotelling T2 control chart in a multivariate manufacturing process
then machine learning algorithms was used to predict the quality of future samples. Four machine
learning algorithms were trained and tested by shifts of different magnitude from the process
average. The performances of the algorithms were compared according to the accuracy and error
rates of the predictions and the most appropriate one was chosen as Multilayer Perceptron.
Keywords: Multivariate Process., Multivariate Control Chart, Machine Learning Algorithm
REFERENCES
[1] H. Hotelling, “Multivariate Quality Control-Illustrated by the Air Testing of Sample
Bombsights,” Techniques of Statistical Analysis, vol. 2(5), 1947, pp. 110-112.
[2] R. L. Mason, N. D. Tracy and J.C. Young, “Decomposition of T2 for Multivariate Control Chart
İnterpretation” J. Qual. Technol. Vol. 27 (2), 1995, pp.99–108.
[3] C. Zhao, C. Wang, L. Hua, X. Liu, Y. Zhang and H. Hu, “Recognition of Control Chart Pattern
Using Improved Supervised Locally Linear Embedding And Support Vector Machine”, Procedia
Engineering vol. 174, 2017, pp. 281 – 288.
[4] M. Zhang and W. Cheng, “Recognition of Mixture Control Chart Pattern Using Multiclass
Support Vector Machine and Genetic Algorithm Based on Statistical and Shape Features”, Mathematical
Problems in Engineering, 2015, pp. 1-10.
[5] S. T. A. Niaki and B. Abbasi, “Fault Diagnosisin Multivariate Control Charts Using Artificial
Neural Network, Quality and Reliability Engineering International, vol. 21, 2005, pp.825-840.
189
[6] M. G. Parra, P. R. Loaiza, “Application of the Multivariate T Control Chart and the Mason–
Tracy–Young Decomposition Procedure to the Study of the Consistency of Impurity Profiles of Drug
Substances”, vol. 16(1), 2003, pp. 127-142.
[7] M. Misra, H. H. Yue, S.J. Q. and C. Ling, “Multivariate Process Monitoring And Fault
Diagnosis By Multi-Scale PCA”, Computers And Chemical Engineering, vol. 26, 2002, pp. 1281-1293.
[8] F. Aparisi, and J. Sanz, “Interpreting the Out-of-Control Signals of Multivariate Control Charts
Employing Neural Networks”, International Journal of Computer and Information Engineering, vol. 61,
2010, pp. 226-230.
[9] L.H. Chen, and T.Y. Wang, “Artificial Neural Networks To Classify Mean Shifts From
Multivariate Χ2 Chart Signals”, Computers And Industrial Engineering., vol.47, 2004, pp.195-205.
[10] He, S., Wang, G.A., Zhang, M., and Cook, D.F., “Multivariate Process Monitoring and Fault
Identification Using Multiple Decision Tree Classifiers”, International Journal of Production Research,
vol.51 (11), 2013, pp. 3355–3371.
[11] E. Alpaydın, “Yapay Öğrenme”, Boğaziçi Üniversitesi Yayınevi, 3. Baskı, 2012, sf. 8-10.
[12] J. Han, M. Kamber, J. Pei, “Data Mining Concepts and Techniques” Third Edition, 2012, pp.
18-20
[13] T. M. Mitchell, “Machine Learning”, McGraw-Hill Science/Engineering/Math, 1997, pp.8185
[14] K-M. Tsai and H-J. Luo, “An Inverse Model for Injection Molding of Optical Lens Using
Artificial Neural Network Coupled with Genetic Algorithm”, Journal of Intelligent Manufacturing, vol.28,
2017, pp. 473–487.
190
Comparison of Two Different Social Groups on Twitter with Network Analysis
Adem AKSAN, Ayşe OĞUZLAR
Department of Econometrics, Uludağ University, TURKEY
Abstract: The social communities created by social media, where people have shown great
interest, have led to the analysis of social network. In this sense, many techniques have been
developed and these techniques have been practised in various fields. By means of the softwares,
developed for investigation of complex network analysis, detailed surveys and research can be
made about social media. In this study, a social network analysis has conducted on the social
media. The study was carried out through the Nodexl program for two different social communities
via Twitter to draw the network graphs and to compare the network analysis results by using ,
“#EytHepBirlikteAnkaraya and #BirlikOlFenerbahçe” According to the results obtained from the
analysis, it is observed that although the EYT members are a more recent social group than
Fenerbahçe supporters, the ties in their network are stronger and the network density of EYT
members is four times more than in Fenerbahçe supporters. In addition, although the hashtags are
addressing different topics, the value of the network characteristics such as clustering and
centralization were found to be similar to each other.
Keywords: Social Media, Social Network Analysis, NODEXL
Giriş
Bilgisayarın icadı ve yaygınlaşmasının ardından internetin ortaya çıkışı ve bilgisayar ile
buluşmasıyla dünyada teknolojik bir devrime sebep oldu. 6 Ağustos 1991 yılında Fransa ve İsviçre
sınırında CERN’de World Wide Web projesi kamuoyuna açıklandı. Tim Berner’s Lee nin
“İnternetteki gezintiniz dünyayı etkiliyor” sözü ile aslında olacakları önceden görmüş gibiydi.
(Gürsakal, 2009) Taşınabilir dizüstü bilgisayarlar akabinde akıllı telefonların ortaya çıkması ile
internet kullanımı dünyada hızla artmaya başladı. İnternette harcanan vakit arttıkça internet
ortamında da çok büyük gelişmeler oldu.
191
Twitter’da ilk tweet 21 Mart 2006’da Jack Dorsey tarafından atıldı. Günümüzde Twitter'da
her dakika ortalama yaklaşık 98 bin tweet atılmaktadır. Atılan tweetler o kadar önem kazanmış ki
ünlü Amerikan TV yıldızı Kim Kardeshian attığı her tweet için markalardan 8 bin dolar aldığı
Ülkemiz basınında geniş yer buldu. Spor dünyası ile ilgili sosyal medya üzerinde yapılan
araştırmada Cristiano Ronaldo’nun Facebook, Instagram ve Twitter’da toplam 262 milyon
takipçisi var ve ünlü futbolcunun sosyal medya sayesinde 500 milyon dolar gelir elde ettiği
söyleniyor (Altan, 2017).
Hatta bireysellikten çıkarak oluşturulan işletme profilleri ile bir markanın, belirli bir grubun
görüşlerini insanlara ulaştırma çabasındadır. Atılan bir tweet beğeni yoluyla binlerce, milyonlarca
insana ulaşabilmektedir. Ücretsiz üye olunan bu ve benzeri siteler sayesinde insanlara en hızlı
ulaşmanın yolu bilinmektedir. Bireyler, sosyal gruplar, şirketlerin yanı sıra siyasi partilerde sosyal
medya aracılığı ile seçmenlerine ulaşmaya çalışmaktadır.
Obama 2008'deki seçim kampanyasın da web teknolojilerini ve sosyal medyayı şüphesiz
en iyi kullanan başkan adayıydı. Kampanyanın yürütülmesi için Facebook’un ortak kurucusu olan
Chris Hughes ile anlaşıldı. Chris Hughes, Barack Obama’nın seçim kampanyasına odaklanmak
için Facebook’taki görevinden istifa etti ve kısa bir sürede kendine bir ekip kurdu. Bu ekiple
birlikte Barack Obama’nın sosyal medya stratejilerini geliştirdi. (Stelter, 2008)
Sosyal Medyada yaşanan bu hızlı gelişmeler, sosyal ağ analizinin sosyal medyaya yönelmesine
neden oldu. Bu çalışmada Twitter üzerinde trend topic olan farklı konularda ki iki hashtag’in
Nodexl programı aracılığıyla ağ grafikleri çizilerek, ağların karşılaştırılması ve yorumlaması
yapılmıştır.
1. Sosyal Ağ Analizi
İnsanoğlu var olduğu günden bu yana sosyal ağların bir parçası olmuştur (Hansen, Shnejderman,
& Smith, 2011). İnsanoğlu akrabalık, dil, din gibi çeşitli nedenlerle birbiriyle sayısız bağlantı
içindedir ve bu bağlar sayesinde oluşan bir ağ dünyası mevcuttur.
Sosyal Ağ Analizi (Social Network Analysis) belirli konularda kişiler, kurumlar, sosyal grup veya
topluluklar arasında ki bağlantıların incelenmesi, tanımlanmasında ve yorumlanmasında
kullanılan kullanılan bir yaklaşım olarak tanımlanmaktadır (John Scott & Carrington, 2011).
192
Bireyler ve kuruluşlar gibi sosyal aktörler arasında ki bilgi akışıyla oluşan bağlantıların
incelenmesi olarak tanımlanmıştır (Himelboim, Smith, Rainie, Shneiderman, & Espina, 2017).
Sosyal ağ: kişilerin ve diğer toplulukların aralarındaki etkileşimi yardımlaşmayı ve etkilerini
gösteren bir yapı olarak da tanımlanmaktadır (Karagöz & Kozak, 2014).
Sosyal ağ analizi, sosyal yapıyı aktörler ve bu aktörleri birbirine bağlayan bir ağ olarak görüp,
etkilerini incelemektedir (Gürsakal, 2009). Bu ağlar köşeler ve düğümlerden oluşur.
Sosyal ağ analizinde odak nokta; sosyal varlıklar arasındaki ilişki ve bu ilişki modellerine
odaklanmasıdır (Karagöz & Kozak, 2014).
Modern sosyal ağ analizi dört temel özelliğe sahiptir. Bunlar:
1. Sosyal ağ analizi, sosyal aktörleri birbirine bağlayan bağlara dayanan yapısal bir sezgiyle
motive edilir,
2. Sistematik ampirik verilere,
3. Büyük ölçüde grafik görüntülerden yararlanır ve
4. Matematiksel ve / veya hesaplamalı modellerin kullanımına dayanır (Freeman L. C.,
2004).
Ağ olarak incelenen sosyal yapı örnekleri, okuldaki çocuklar arasındaki dostluk, sosyal seçkinlerin
üyeleri arasındaki aile ilişkileri, şirketlerin ortak yönetim kurulu üyeleri, ülkeler arasındaki ticari
ilişkiler ve web siteleri arasındaki köprülerdir (Barabási, 2002).
Şekil 1. İnternet Ağı (Barabási, 2002)
193
Şekil 2. Hollywood Aktör Ağı (Barabási, 2002)
Şekil 3. Protein Etkileşim Ağı
Yukarıda gösterilen üç farklı ağlardan;
1’nci ağ; yönlendiricilerin (uzman bilgisayarların) birbirine bağlandığı küçük bir internet
ağını,
2’nci ağ; filmde oynarlarsa iki aktörün birbirine bağlandığı Hollywood aktör ağı,
3’ncü ağ (c) Hücrede birbirlerine bağlanabilecekleri deneysel kanıt varsa, iki proteinin
bağlandığı bir protein-protein etkileşim ağını göstermektedir.
Düğümlerin ve bağlantıların niteliği farklı olsa da bu ağlar, 4 düğüm ve 4 bağlantıdan
oluşan aynı grafiksel gösterime sahiptir (Barabási, 2002).
194
Sosyal ağ analizini ağlar olarak kavramsallaştırılabilecek verilere uygulamanın her zaman
yararlı veya gerekli olduğunu söylemek doğru değildir. Örneğin, eğer bir araştırmacı, bir kişinin
yardım için başvurabileceği kişi sayısını bilmekle ilgileniyorsa, bağların yapısı yerine bağların
sayısı önemlidir ve ağ analizine gerek yoktur. Ağ analizinin uygulanabilmesi için, sosyolojiden ya
da diğer sosyal ve davranış bilimlerinden gelen teori, bağların yapısının ağ üyelerinin davranışları,
görüşleri ya da sosyal konumlarıyla bağlantılı olduğuna inanmak için sebepler vermelidir
(Barabási, 2002).
Son yıllarda geliştirilen teknikler sayesinde sosyal ağ analizi değişik alanlarda da
uygulanmaya başlamıştır. Karmaşık ağ analizlerinin incelenmesi için geliştirilen yazılımlar
sayesinde ağlar konusunda detaylı araştırma ve incelemeler yapılabilmektedir.
Twitter, Facebook ve Instagram gibi sosyal medya sitelerinde gösterilen yoğun ilgi sosyal
medya üzerinde Sosyal Ağ Analizi çalışmalarına ivme kazandırmıştır. 2018 yılı global digital
raporu verilerine göre (DIGITAL AROUND THE WORLD IN 2018) dünyada 3 milyar 196
milyon kişi sosyal medya kullanmaktadır. Ortalama bir internet kullanıcının internette geçirdiği
süre ise günde ortalama 6 saat (Kemp, 2018).
Milyarlarca insanın günde ortalama 6 saat vakit geçirdiği ortamda şirketler, siyasi parti ve liderler,
popüler kişi ve gruplar sosyal medya sayesinde tanıtımlarını yaparak hedef kitlelerine ulaşmaya
çalışmaktadırlar.
2.
Nodexl
Sosyal Medya Araştırma Vakfı (The Social Media Research Foundation) tarafından Temmuz
2008’de kullanıma sunulan Nodexl programı Microsoft Excel’e eklenti olarak eklenip Twitter,
Facebook, Youtube ve Flickr'dan karmaşık sosyal ağları toplamanızı, analiz etmenizi ve
görselleştirilmesini sağlamaktadır. ( The Social Media Research Foundation)
Kısaca NodeXL sosyal medyadan veri toplayarak ağ görselini ve raporunu oluşturur. Basit
filtreleme ve ekran özellikleri ile ağlarda önemli vurguları yapmak için kullanılır. NodeXL de
işlemler 5 adımda yürütülmektedir.
1. Çeşitli kaynaklardan veri toplama
2. Veri depolama
195
3. Veri Analizi
4. Görselleştirme
5. Yayınlama ( The Social Media Research Foundation)
Nodexl ile yapılan çalışmaları incelediğimizde; Himelboim ve arkadaşları (2017), Twitter
üzerindeki ağları sınıflandırarak bilgi akışına göre modellemişlerdir (Himelboim, Smith, Rainie,
Shneiderman, & Espina, 2017).
Pew Araştırma Merkezi ve Sosyal Medya Araştırma Vakfı tarafından yapılan çalışmada
binlerce Twitter görüşmesinin özel bir analizi ile Twitter'da gerçekleşen konuşma ve sosyal
yapılara altı farklı kalıbın olduğunu ortaya çıkarmışlardır (Marc Smith, 2014).
IEEE Üçüncü Uluslararası Sosyal Bilgi İşlem Konferansında Rodriges ve arkadaşları
(2011), Sosyal ağlarda yer alan topluluklarda ki bireylerin hangi kategorilere göre (yaş, cinsiyet,
meslek, coğrafi konum vb.) birbirleriyle etkileşim içinde olduğunu incelemişlerdir (Eduarda
Mendes Rodrigues, 2011).
Hibeilboim ve arkadaşları (2009), forum sitelerinde yapılan politik tartışmaları sonucu
ortaya çıkan sosyal ağlarda tartışmayı başlatan, büyüten ve tartışmanın içeriğini belirleyen ve
değiştiren kullanıcılar(katalizör) tespit edilmeye çalışılmıştır (Himelboim I. E., 2009).
Araştırmamızda sosyal medya siteleri içerisinde yer alan Twitter üzerinden yapılacaktır.
Twitter 2006 yılında kullanılmaya başlanılmış olup 2018 yılı global digital raporuna göre
Twitter’da kullanıcı sayısı 330 milyondur (Kemp, 2018).
Bir sosyal ağ perspektifinden sosyal medya araştırması, odağı bireysel özelliklerden sosyal
varlıklar arasındaki ilişkisel bağlara kaydırır. (Bruns & Stieglitz, 2013) Bu bağların koleksiyonları,
ortaya çıkan desenler veya ağ motifleri halinde toplanır. Sosyal ağ sitelerinde, kullanıcılar
kendileriyle bağlantı kurdukları veya bilgi paylaştıkları zaman diğer kullanıcılarla etkileşim
kurarak ağ oluştururlar. Twitter'da sosyal ağlar, kullanıcılardan bahsettiklerinde ve birbirlerine
cevap verdiklerinde ve diğer kullanıcılarla kurdukları bağlantılardan oluşur (Hansen,
Shnejderman, & Smith, 2011).
196
2.1. Ağlara ilişkin Ölçümler
Düğümler (Vertices), aracılar, varlıklar veya öğeler olarak da adlandırılan tepe noktaları birçok
şeyi temsil edebilir. Genellikle insanları veya çalışma grupları, takımlar, organizasyonlar,
kurumlar, eyaletler ve hatta ülkeler gibi sosyal yapıları temsil ederler. Diğer zamanlarda, web
sayfaları, anahtar kelime etiketleri veya videolar gibi içerikleri temsil ederler. Fiziksel veya sanal
konumları veya olayları bile temsil edebilirler.
Bağlantılar (Edges), bağlar, bağlantılar ve ilişkiler olarak da bilinen kenarlar, ağların yapı
taşlarıdır. Bir kenar iki köşeyi birbirine bağlar. Kenarlar, yakınlık, iş birlikleri, akrabalık, dostluk,
ticari ortaklıklar, alıntılar, yatırımlar, köprü oluşturma, işlemler ve paylaşılan özellikler gibi birçok
farklı ilişkiyi temsil edebilir. Resmi bir statü statüsüne sahipse, katılımcılar tarafından tanınıyorsa
veya aralarındaki değiş tokuş veya etkileşimle gözlemlenirse, bir bağın var olduğu söylenebilir.
Bir bağ, iki varlık arasındaki herhangi bir ilişki veya bağlantı şeklidir.
Bir diğer ölçüm ise “yoğunluk” (density) dur. Yoğunluk düğümler arasındaki bağlantıların
yoğunluğunu veya seyrekliğini göstermektedir. Bir ağın yoğunluğu, mevcut bağlantı sayısının
mümkün olan maksimum bağlantı sayısına oranı olarak gösterilmektedir (Wasserman & Faust,
1994).
“Çap” (diameter) ağdaki bütün düğüm çiftlerinin arasındaki en kısa patikaların en
uzunudur. Çap ne kadar kısa olursa o ağda bilgi o kadar hızlı yayılır. (Gürsakal, 2009)
Öz döngüler (self-loops) kendi içinde bağlantı olmayan düğümleri gösterir. Kullanıcının
ağ içerisinde kendi kendine bağlı olduğunu göstermektedir. Marco Bastos Şubat 2014’ te
yayınladığı makalesinde “öz döngüler ağların yapısı ile yakından ilişkilidir ve kullanıcılar hastac’
leri kendi profilin de yayınlama eğilimindedir.” Öz döngülerin ağlarda ki önemine vurgu yapmıştır
(Bastos, 2014).
2.1.1 Ağlarda Kümelenme (Clusters ve Modularite (Modularity)
Pek çok ağın ortak bir özelliği, aynı üçüncü düğüme bağlı olan iki düğümün de birbiriyle bağlanma
olasılığının daha yüksek olduğu kümeleme veya ağ geçirgenliğidir. Basitçe söylemek gerekirse,
belirli bir topluluktan yapılacak rastgele seçimden ise arkadaşlarınızdan ikisinin birbirlerini tanıma
olasılığı daha yüksektir. Bir ağ büyüdükçe, kümeler daha büyük bir ağ içindeki düğüm alt kümeleri
197
daha fazla birbirine bağlanır, yani yoğunlaşır bu kümeler arasındaki bağlantılar daha az yoğundur
(Newman & Girvan, 2004).
Twitter’da kullanıcılar birbirlerini takip ederek bilgi akışı için yol oluşturmuş olurlar. Bu
yöntem ile kullanıcıların ilgi alanlarına göre gruplar oluşmuş olur. Ortaya çıkan bu gruplar bilgi
akışının sosyal sınırını (Himelboim, Smith, Rainie, Shneiderman, & Espina, 2017) belirlemiş olur.
Oluşan bu kümelerde bilgi serbestçe akarken kümeler arasında ki bilgi akışı, kümeler arasındaki
bağlantılarla sağlanmaktadır.
Sosyal medya ağlarında ki kümelenmenin önemini Conver ve arkadaşları (2011), siyasi
düşünceleri benzer kullanıcıların retweet kalıplarını ortaya çıkarmışlardır. (M. D. Conover, 2011)
Ağ kümelerinin bilgi akışında ki rolü daha sonra Rodrigez ve arkadaşlarının (2012),
çalışması ile de gösterildi (Rodrigez, Leskovec, & Krause, 2012).
Kümelenme katsayısı (Clustering Coefficient)
Kümelenme katsayısı, yoğunluk(density) ölçüsüne benzerdir. Eğer ağda kullanıcılar birbirini
tanıyorsa yüksek, bir kümelenme katsayısına sahip olursunuz. Eğer “arkadaşlarınız” (değiştirir)
birbirinizi tanımıyorsa, düşük kümelenme katsayısına sahip olursunuz. Kümelenme katsayısı,
kullanıcıların başkalarıyla bağlantı kurma biçimlerine ve içinde bulundukları ortamlara bağlı
olarak farklı ölçütlere sahip olabilir.
Şekil 4. Kümelenmiş ağ örneği (Kaiser, Görner, & Hilgetag, 2007)
Modülarite (Modularity) Ağın yapısının modülerliği, birbirine bağlı alt kümelere bölünmüş
bir ağ olan kümelenme kalitesinin bir ölçüsüdür. (Newman & Girvan, 2004) Modülerlik,
198
kümelenmiş ve birleştirilmiş yapılara sahip ağlar arasında ayrım yapan kümelerin birbirinden
ayrılma derecesini (0'dan 1'e kadar bir değer aralığı) yakalar. Ayrıca, iki çok farklı türde yoğun
ağların ayırt edilmesine yardımcı olur. (Himelboim, Smith, Rainie, Shneiderman, & Espina, 2017)
Modülerlik, kümeler arasında bölünmenin ne kadar iyi olduğunu, kümeler içinde birçok bağlantı
olduğunu ve sadece birkaçının olduğunu ölçmektedir. Modülerlik değerleri 0 ile 1 arasında
değişmektedir. Modülerlik değeri ne kadar yüksekse, kümeler o kadar belirgin veya ayrıktır, yani
kümeler birbiriyle daha az bağlantılıdır (Himelboim, Smith, Rainie, Shneiderman, & Espina,
2017).
2.1.2 Ağlarda Merkezileştirme (Centralization)
Herhangi bir ağın merkezileşmesi, en merkezi düğümünün diğer tüm düğümlerin merkeziyetiyle
ne kadar merkezi olduğuna dair bir ölçüdür (Freeman L. C., 1979).
Bir ağda bilgi akışı kullanıcılar arasında ki bağlantıların dışında ağ yapısının hiyerarşik
veya eşitlikçi olmasıyla ilgilidir. Bir veya birkaç sosyal aktörün sayısız bağlantılarının olduğu bir
ağdaki bilgi akışı bu aktörlere bağlı olarak daha hiyerarşik iken, bağlantı sayılarının azalarak daha
fazla sosyal aktörlere yayıldığı ağlar daha merkezidir şeklinde tanımlayabiliriz (Himelboim,
Smith, Rainie, Shneiderman, & Espina, 2017).
Van Den Bos (2006), Hollanda'daki İranlıların web siteleri arasında köprüler tarafından
oluşturulan çevrimiçi topluluk yapısının, bu ağda odaklanan birçok nokta ile düşük merkezileşme
gösterdiğini ortaya koymuştur.
Woo-young ve Park (2012), yaptıkları çalışmada ABD yer alan haber bloglarının
bulunduğu ağın merkezi bir ağ olduğunu tespit etmiştir (Woo-young & Park, 2012).
Merkezilik bir tür popülerlik ölçütü olarak düşünülebilir. Derece, belirli bir tepe noktasına
bağlı toplam kenar sayısının ölçüsüdür. Yönlendirilmiş ağlar için iki derece ölçüsü vardır. Bir
köşeye içe dönük bağlantıların sayısı girdi derecesi (Input degree), bir köşeden kaynaklanan ve
diğer köşelere dışa dönük bağlantılar ise çıktı derecesi (output degree) olarak bilinmektedir
(Gürsakal, 2009).
Arasındalık Merkeziliği (Betweenness Centrality), bir mesafe ölçüsüdür. Yol kavramı
ağların araştırılmasında merkezi bir öneme sahiptir. Belki de bir ağdaki herhangi iki kişi hakkında
sorulacak en doğal sorulardan biri “Ne kadar uzaktalar? “Bu mesafe basitçe ölçülür: komşusu
199
olmayan insanlar arasındaki mesafe, en az sayıda komşu-komşu atlama sayısı birinden diğerine
ölçülür. Örneğin, sizin komşunuz olmayan, ancak komşularınızın komşusu olan insanlar, sizden 2
uzakta. İki kişi arasındaki en kısa yol “jeodezik mesafe” olarak adlandırılır ve birçok merkezi
ölçümde kullanılır (Himelboim I. E., 2009).
Yakınlık merkezliliği (Closeness centrality), bir düğüm ile ağdaki diğer her düğüm
arasındaki ortalama mesafeyi ifade etmektedir. Düğümlerin yalnızca mesajlarını yalnızca mevcut
bağlantılarına iletebileceğini veya verebileceğini varsayarsak, düşük bir yakınlık merkezi olması,
bir kişinin doğrudan bağlı olduğu veya ağdaki çoğu kişiden daha hızlı bir şekilde diğer kişilere
ulaşabildiği anlamına gelmektedir. Örneğin; şehirde yaşayan insanların şehir merkezine ortalama
mesafesini yakınlık merkezi ölçüsü olarak düşünebiliriz (Denny, 2014).
Özvektör merkeziyeti: Bir aktörün diğer iyi aktörlere bağlanma derecesini ölçer. Özvektör
merkeziyetçiliği, daha sofistike bir merkeziyet anlayışıdır. Az sayıda bağlantıya sahip bir kişi, eğer
bu az bağlantıların kendisi çok iyi bağlanmışsa, çok yüksek öz vektörlük merkeziyetine sahip
olabilir. Öz vektör merkezliliği, bağlantıların değişken bir değere sahip olmasını sağlar, böylece
bazı köşelere bağlanmak diğerlerine bağlanmaktan daha fazla yarar sağlar.
3. Uygulama
Ağ grafikleri, bir etkinliğin temelini oluşturan sosyal yapıyı anlamak, bir olayla ilgili kilit kişileri
tanımlamak, etkinliğin etrafındaki sohbeti haritalamak ve zaman içinde izlemek ve ilgili olayları
karşılaştırmak için kullanılabilir (Hansen, Shnejderman, & Smith, 2011).
Son dönemlerde emeklilikte yaşa takılanlar (EYT), gündemde ve sosyal medyayı çok aktif
kullanılıyorlar. Bu grubun sosyal yapısını anlamak ve kilit kullanıcıları tanımlamak ve EYT’lilere
göre daha eski bir sosyal grup ile karşılaştırmasını yapmak amacıyla;
09.02.2019-10.02.2019 tarihleri arasında Twitter’da trend topic olmuş hashtagler arasından
#EytHepBirlikteAnkaraya ve #BirlikOlFenerbahçe seçilmiş ve Nodexl aracılığıyla ağ grafikleri
çizilmiştir. 1Şekil 5 ve Şekil 6’daki ağ grafikleri incelendiğinde 2.ağda kümelenmenin daha fazla
olduğu ve gruplar arasında ki bağlantıların 1.ağa göre daha az olduğu gözlemlenmiştir.
200
Nodexl aracılığıyla elde edilen grup istatistiklerine bakıldığında 1.ağda 72, 2.ağda 742,
grup olduğunu tespit etmiştir. Çizilen bu iki ağ grafiği de Newman ve Girvan’ın 2004 yılındaki
çalışmasında kümelenme hakkında yaptığı tanımı destekler niteliktedir.1 2.ağda ki bilgi akışı
grupların kendi içinde ki bağlantılarla sağlanırken, 1.ağda ise bu bağlantılar daha etkin kullanıcılar
sayesinde gerçekleşmektedir.
Ağlarda ki düğümler genellikle dairesel şekilde gösterilir. Bu çalışmada ağdaki en aktif
düğümleri, bir nevi ağlardaki başrolleri, popüler kişileri öne çıkarmak amacıyla, ilk 10 düğümü
profil resimleri ile gösterilmiştir. Burada ilk 10’u belirlemek için en fazla bağlantı alan (Input)
düğüme göre sıralama yapılarak ağ grafikleri çizilmiştir.
201
Şekil 5. #EytHepBirlikteAnkaraya Kümelenmiş Ağ Grafiği
202
Şekil 6. #BirlikOlFenerbahçe Kümelenmiş Ağ Grafiği
203
Tablo 1. Grafik Ölçümleri (Graph Metrics)
Grafik Ölçümleri
Vertices(düğüm)
1.Ağ
2.Ağ
#EytHepBirlikteAnkaraya #BirlikOlFenerbahçe
1072
1364
Unique Edges (benzersiz
kenarlar)
Edges With Duplicates
(tekrarlanan kenarlar)
2785
1725
403
342
Total Edges (toplam kenar)
3188
2067
Self-Loops (öz döngü)
129
768
Reciprocated Vertex Pair Ratio
(karşılıklı düğüm çiftlerinin
oranı)
0,000345066
0,004966887
Reciprocated Edge Ratio
(karşılıklı kenar oranı)
0,000689893
0,009884679
Connected Components
(birbiriyle bağlı bağlantılar)
57
578
Single-Vertex Connected
Components (tek bağlantı)
41
525
Maximum Vertices in a
Connected Component
(maksimum düğüm sayısı)
994
704
Maximum Edges in a Connected
Component (maksimum kenar
sayısı)
3125
1338
Maximum Geodesic Distance
(Diameter)(çap)
9
10
Average Geodesic Distance
(ortalama çap)
3,833291
3,868626
Graph Density (yoğunluk)
0,002478852
0,000652993
Modularity (modularite)
0,42884
0,568192
202
1.ağımızda toplam 1072, 2. Ağımızda toplam 1364 düğüm bulunmakta olup, 2.ağ daha
büyük bir ağdır. 1. Ağdaki kenar sayısı (3188), yani ağ içinde ki bağlantı sayısı 2.ağa (2067)
göre daha fazladır. 2.ağda öz döngü sayısı 1.ağa göre daha fazla olup 768 adet bağlantının diğer
bağlantılarla ilişkisi olmayıp sadece kendi profillerinde paylaşımda bulunulmuştur.
2.ağda karşılıklı düğüm oranı (0,009) 1.ağa (0,0003) göre büyük olup 2.ağda ki
düğümlerin etkileşimi, daha büyüktür. Aynı yorumu karşılıklı kenar oranı içinde yapabiliriz. 2
ağdaki bağlantıların birbirleriyle etkileşimi daha büyüktür. Düğüm ve kenar oranlarının
yorumuna göre birbirleriyle bağlı bağlantı sayısında 2.ağda daha fazla olması beklenilmelidir.
Tablo 1’i incelediğimizde 2.ağdaki birbirleriyle bağlı bağlantı sayısı (578) daha fazladır.
Tek bağlantılara baktığımızda 2.ağdaki (525) bağlantı sayısı daha fazladır. 1.ağdaki
Birbirine bağlı maksimum düğüm (994) ve maksimum kenar (3125) sayısı 2.ağa göre daha
fazladır.
Maksimum çap 2.ağda (10) daha büyüktür. Ortalama çaplara baktığımızda ise tüm
düğümler arasındaki ortalama uzaklıklar (1.ağ 3,83-2.ağ 3,86) birbirlerine yakın değerdedir. Bu
değer sıfıra yaklaştıkça ağdaki bilgi daha hızlı yayılır.
1.ağın yoğunluğu (0,0024), 2.ağa (0,0006) göre daha fazla olup bu değer 1’e yaklaştıkça
ağdaki bilgi akışı ve etkileşim artar. Modularite değerlerini karşılaştırdığımız da 2. Ağın
modularite değeri (0,56) 1’e daha yakın olup, 2.ağdaki kümelenme daha kalitelidir. Bunun
anlamı 2.ağda kümelenme daha belirgin, kümeler arasında ki bağlantılar daha azdır. Ağlarda ki
gruplar kümelenme işlemi sonucu oluşur. Kümeleme işlemi bağlantılar ne kadar az ise o kadar
belirgin olur. Buna göre Twitter’da kullanıcıların birbirlerine çok bağlı olduğu ağlar daha az
modüler veya kullanıcıların birbirine daha az bağlı olduğu ağlar daha fazla modülerdir
diyebiliriz (Newman & Girvan, 2004).
203
Tablo2. Arasındalık Merkeziliği Değerleri (Betweenness Centrality)
1.Ağ #EytHepBirlikteAnkaraya
2.Ağ #BirlikOlFenerbahçe
1000
1500
Frequency
Frequency
800
600
400
500
0
200
0
1000
Betweenness Centrality
Betweenness Centrality
Minimum Betweenness
Centrality
Maximum Betweenness
Centrality
Average Betweenness
Centrality
Median Betweenness
Centrality
Minimum Betweenness
Centrality
Maximum Betweenness
Centrality
Average Betweenness
Centrality
Median Betweenness
Centrality
0,000
153965,372
2873,237
0,000
195580,697
1045,268
0,000
58,551
Arasındalık merkeziliği ölçülerine baktığımızda iki düğüm arasındaki en kısa mesafe
2.ağda daha yüksek iken, tüm düğümler arasından iki düğüm arasında ki ortalama mesafe 2.
Ağda daha düşüktür. 2.ağda kullanıcılar arasında ki komşuluk bağları daha iyidir. Bu ölçüm
değeri sıfıra yaklaştıkça komşuluk bağları artar.
Tablo 3. Yakınlık Merkeziliği Değerleri (Closeness Centrality)
204
Yakınlık ölçüsünü bir düğümün diğer tüm düğümlere ortalama mesafesi olarak
tanımlamıştık. İki ağda ki ortalama ölçüm değerlerine baktığımızda 1.ağda ortalama değer
(0,028) daha düşük olup 1.ağda ki bir kullanıcı 2.ağda ki bir kullanıcıya göre daha hızlı bir
şekilde diğer kullanıcılara ulaşabilmektedir. EYT’lilerin sosyal medyada bu kadar etkin
olmalarının sebebi de zaten bu olsa gerek.
Tablo 4. Öz vektör Merkeziliği Değerleri (Eigenvector Centrality)
2.Ağ #BirlikOlFenerbahçe
1000
Frequency
Frequency
1.Ağ #EytHepBirlikteAnkaraya
0
2000
0
Eigenvector Centrality
Eigenvector Centrality
Minimum Eigenvector Centrality
0,000
Maximum Eigenvector Centrality
Average Eigenvector Centrality
Median Eigenvector Centrality
0,015
0,001
0,000
Minimum Eigenvector Centrality
Maximum Eigenvector Centrality
Average Eigenvector Centrality
Median Eigenvector Centrality
0,000
0,030
0,001
0,000
Özvektör merkeziliği ait ortalama ölçüm değerine baktığımızda ise değerler birbirine
eşit. Her iki ağda da düğümlerin birbirine ortalama aynı derece ile bağlanıyor. Her iki ağda da
düğümler arasındaki bağlar güçlü.
Yukarıda Merkezilik ölçüm sonuçlarına ilişkin değerlendirmeleri yaptık. 2 ağımızda da
derece dağılımına baktığımız kuvvet yasası geçerli gibi. Az sayıda kullanıcının etkinliği
ağlarımızda mevcut.
Tablo5. Kümelenme Katsayısı (Clustering Coefficient)
1.Ağ #EytHepBirlikteAnkaraya
2.Ağ #BirlikOlFenerbahçe
1500
Frequency
Frequency
1500
1000
500
0
1000
500
0
Clustering Coefficient
Clustering Coefficient
205
Minimum Clustering Coefficient
0,000
Maximum Clustering Coefficient
1,000
Average Clustering Coefficient
0,009
Median Clustering Coefficient
0,000
Ortalama kümelenme katsayısına baktığımızda ise değerler birbirine yakın kümelenme
katsayısı sıfırdan uzaklaştıkça kullanıcıların birbirini daha fazla tanıdığı yorumlanabilir. Bu
katsayıya göre 1.ağ azda olsa 2.ağa göre birbirini daha fazla tanımaktadır.
Sonuç
Bilgi çok hızlı ve kontrolsüz bir şekilde yayılabiliyor. Sosyal medya, sosyal aktörleri ve bu
aktörlerden oluşan toplulukları önemli hale getirdi. Bizimde amacımız sosyal medya da
topluluklar aracılığıyla gerçekleştirilen bilgi akışını kümelenmiş ağ grafikleri ile göstermek ve
ve sosyal medyada ki etkin grupların oluşturduğu trend topicleri ağ grafikleri ve ölçümleri ile
ağları karşılaştırmak ve doğru bir şekilde yorumlayabilmekti. Bunun için ağların
görselleştirilmesi ve ölçümleri konusunda Nodexl programı kullanıldı.
İki ağda farklı konulardan oluşan hashtaglerden oluşsa da ağların ölçüm değerlerine
baktığımızda birbirine yakın değerler elde edildi. Trend topic olmaları bu ağların benzer
sonuçlar göstereceği düşünülse de ağları birbirinden ayıran farklı dinamikler olduğunu tespit
edildi. 2.ağda ki grup sayısı 1.ağa göre daha fazla olsa da (1.ağ 72, 2.ağ 742) kümelenme
katsayıları birbirine çok yakın değerler (1.ağ, 0,010, 2.ağ 0,009) elde edilmiştir. 1.ağda ki
emeklilikte ki yaşa takılanlar, 1.ağda ki Fenerbahçelilere göre daha yeni grup olsalar da
aralarındaki sıkı etkileşim nedeniyle birbirlerini daha fazla tanımaktadırlar. Modularite
değerlerine göre ise 2.ağda kümelenmenin (0,56), 1.ağa göre daha kaliteli ve belirgin olduğunu
bize gösterdi. 1.ağın yoğunluğu 2.ağa göre 4 kat daha fazladır (0,0024-0,006). 1 ağda bağlar
daha sıkı ve bilgi akışı 2.ağa göre 4 kat daha hızlıdır.
Sosyal medyada Fenerbahçeliler EYT grubuna göre çok daha eski sosyal topluluk
olmasına rağmen yapılan ölçüm sonucunda EYT’lilere ilişkin ağda ki bilgi akışının
hızı(yoğunluk), modularite gibi değerlerin daha yüksek olduğu, kümelenme katsayısı,
merkezileşme ve çap gibi değerlerinde birbirine çok yakın değerlere sahip olduğu tespit
edilmiştir.
206
Sosyal medya yadsınamaz bir gerçek haline geldi. Bu analiz sosyal medya aracılığı ile
tanıtım yapan kuruluşların doğru stratejiler geliştirmesi için kullanışlı olduğu gibi, Sosyal
medyada ki başarılarını da bu sayede değerlendirmeye alabilirler. Twitter üzerinden yapılan bu
çalışma Nodexl programı sayesinde Facebook, Instagram, Youtube vb. sosyal medya
kanallarındaki ağlar içinde uygulanabilir ve analiz edilebilir.
REFERENCES
[1] The Social Media Research Foundation. Smr foundation. Ocak 28, 2019 tarihinde The Social
Media Research Foundation: https://www.smrfoundation.org/ adresinden alındı.
[2]
Altan,
S.
www.pazarlamasyon.com:
(2017,
Mart
15).
Pazarlamasyon.
Ocak
24,
2019
tarihinde
https://pazarlamasyon.com/cristiano-ronaldonun-sosyal-medya-
hesaplarindan-bir-yilda-500-milyon-dolar-gelir-elde-edildi/ adresinden alındı
[3] Barabási, A.-L. (2002). Linked: The New Science of Networks. UK: Perseus Publication.
[4] Bastos, M. (2014, Şubat 28). Hastac. Ocak 30, 2019 tarihinde www.hastac.org/:
https://www.hastac.org/blogs/herrcafe/2014/02/28/self-loops-and-network-awareness adresinden alındı
[5] Bruns, A., & Stieglitz, S. (2013). Towards More Systematic Twitter Analysis: Metrics for
Tweeting Activities. International Journal of Social Research Methodolog, s. 91-108.
[6] Denny, M. (2014, Eylül 26). Social Network Analysis. Unıversity of Massachusetts
Amherst, Institute For Social Science Resarch, s. 1-20.
[7] Eduarda Mendes Rodrigues, N. M.-F. (2011). Group-in-a-box Layout for Multi-faceted
Analysis of Communities. IEEE Third International Conference on Social Computing (s. 9-11). Boston:
IEEE.
[8] Freeman, L. C. (2004). The Development of Social Network Analysis: A Study in The
Sociology Of Science. Vancouver: Empirical Press.
[9] Gürsakal, N. (2009). Sosyal Ağ Analizi. Bursa: Dora Yayınları.
[10] Hansen, D., Shnejderman, B., & Smith, M. (2011). Analyzing Social Media Networks With
Nodexl: Insights From A Connected World. Burlington MA: Morgan Kaufmann.
[11] Himelboim, I. E. (2009). Discussion Catalysts in Online Political Discussions: Content
İmporters And Conversation Starters. Journal of Computer-Mediated Communication, 771-789.
207
[12] Himelboim, I., Smith, M., Rainie, L., Shneiderman, B., & Espina, C. (2017, Ocak-Mart).
Classifying Twitter Topic-Networks Using Social Network Analysis. Social Media + Society, s. 1-13.
[13] John Scott, & Carrington, P. (2011). The SAGE Handbook of Social Network Analysis.
London: Sage Publications Ltd.
[14] Kaiser, M., Görner, M., & Hilgetag, C. C. (2007, Mayıs 2). Criticality of Spreading
Dynamics in Hierarchical cluster Networks without Inhibition. New Journal of Physics, s. 1-13.
[15] Karagöz, D., & Kozak, N. (2014). Anatolia Turizm Araştırmaları Dergisi’nin Bibliyometrik
Analizi: Araştırma Konuları ve Kurumlar Arası İş Birliğinin Sosyal Ağ Analizi ile İncelenmesi. Türk
Kütüphaneciliği, 47-61.
[16] Kemp, S. (2018, Ocak 30). DIGITAL IN 2018: WORLD’S INTERNET USERS PASS
THE
4
BILLION
MARK.
Ocak
24,
2019
tarihinde
https://wearesocial.com:
https://wearesocial.com/blog/2018/01/global-digital-report-2018 adresinden alındı
[17] M. D. Conover, J. R. (2011). Political Polarization on Twitter. Proceedings of the Fifth
International AAAI Conference on Weblogs and Social Media (s. 89-96). Bloomington: ICWSM.
[18] Marc Smith, L. R. (2014, Şubat 20). Mapping Twitter Topic Networks: From Polarized
Crowds to Community Clusters. USA.
[19] Newman, M. E., & Girvan, M. (2004). Finding And Evaluating Community Structure in
Networks. PHYSICAL REVIEW, 1-15.
[20] Rodrigez, M. G., Leskovec, J., & Krause, A. (2012). Inferring Networks of Diffusion and
Influence. Transactions on Knowledge Discovery from Data, 21/1-36.
[21] Stelter, B. (2008, Temmuz 7). The New York Times. Ocak 24, 2018 tarihinde
www.nytimes.com: https://www.nytimes.com/2008/07/07/technology/07hughes.html adresinden alındı
[22] Wasserman, S., & Faust, K. (1994). Social Network Analysis: Methods and Applications.
Cambridge: Cambridge University Press.
[23] Woo-young, C., & Park, H. W. (2012). The Network Structure of the Korean Blogosphere.
Journal of Computer-Mediated Communication, 216-230.
208
Analysis of Earthquake Awareness in Education By Data Mining
Ömer KIVRAK, Filiz ERSÖZ
Department of Industrial Engineering, Karabük University, TURKEY
Abstract: Turkey and the world a lot of earthquakes that have occurred and will continue to
come. Due to the earthquake, many people have caused their lives and damage to their shelters.
Starting from a young age earthquake in Turkey to create awareness in every age group,
observed and loss of life and property damage in the earthquake is one of the lowest levels to
minimize the road. Therefore, it is very important to be able to use techniques and systems that
can analyze a large number of data sets. The process of converting these raw data into
information or meaning can be done by data mining. The aim of this study is to investigate the
effects of the education given to the secondary and high school students on the students and the
earthquake awareness. The research sample consisted of 14 middle school and 11 high school
students, who were randomly selected in Karabük and districts. The questionnaire developed
by the researcher was used as the data collection tool. In the scope of the research, 1165 students
were given questionnaires before and after 80 minutes of training. In this study, the seismic
awareness of the students' education on the students before and after the training was
investigated by clustering analysis.
Keywords: Earthquake, Earthquake Awareness, Data mining, Clustering analysis.
REFERENCES
[1] R. Kasap and Ü. Gürlen, “Deprem Magnitüdleri İçin Tekrarlanma Yıllarının Elde Edilmesi:
Marmara Bölgesi Örneği,” Doğuş Üniversitesi Dergisi, cilt 4, no. 2, pp. 157-166, 2003.
[2] S. Savaş, N. Topaloğlu and M. Yılmaz, “Veri Madenciliği ve Türkiye’deki Uygulama
Örnekleri,” İstanbul Ticaret Üniversitesi, Fen Bilimleri Dergisi, cilt 11, no. 21, pp. 1-23, 2012.
[3] O. İnan, Veri Madenciliği, KONYA: Selçuk Üniversitesi, Fen Bilimleri Enstitüsü, 2003.
[4] B. Thuarisingham, Web Data Mining And Applications in Business Intelligence And
Counter Terrorism, Boca Raton, Fl, Usa, 2003.
209
[5] Z. Z. Bakır, “İnsani ve Sosyal Araştırma Merkezi,” İnsani ve Sosyal Araştırma Merkezi,
Http://İnsamer.Com/Tr/Dunyanin-Afet-Tablosu-Ve-Cozum-Onerileri_380.Html 2018.
[6] C. Geray, “Türkiye‘de Yıkım (Afet) Olayları Karşısında Önlemler Vv Örgütlenmeler,”
Amme İdaresi Dergisi, Cilt 7, No. 2, Pp. 91-114, 1977.
[7] H. Yavaş, Doğal Afetler Yönüyle Türkiye’de Belediyelerde Kriz Yönetimi, Ankara: Orion
Yayınevi, 2005.
[8] A. Yılmaz, Türk Kamu Yönetiminin Sorun Alanlarından Biri Olarak Afet Yönetimi, Ankara:
Pegem Yayıncılık, 2005.
[9] Ö. Kıvrak And S. Kardeş, Kuzey Anadolu Fay Hattı Üzerinde Karabük İli ve Çevresinde
Oluşan Depremlerin Veri Madenciliği Kullanılarak İncelenmesi, Karabük, 2014.
[10] F. Ersöz, Veri Madenciliği Teknikleri ve Uygulamaları, Seçkin Kitabevi, 2019.
[11] S. Piramuthu, “Evaluating Feature Selection Methods For Learning in Data Mining
Applications,” %1 İçinde Thiry-First Hawai International Conference On System Sciences, Kohala
Coast Hawaii Usa, 1998.
[12] S. Özekes, “Veri Madenciliği Modelleri ve Uygulama Alanları” İstanbul Ticaret
Üniversitesi Feb Bilimleri Dergisi, Cilt 2, No. 3, Pp. 65-82, 2003.
[13]
S.
Ilgaz,
“Eğitimde
Farkındalık,”
[Ttps://Www.Kucukdahi.Com/Blog/Egitimde-
Farkindalik. 30 Nisan 2019.
[14] A. M. Bauer, J. A. Quas And W. T. Boyce, “Associations Between Physiological Reactivity
And Children's Behavior: Advantages Of A Multisystem Approach,” Journal Of Developmental And
Behavioral Pediatrics, Cilt 23, P. 102–113, 2002.
[15] R. Mccraty, N. Atkinson, D. Tomasino, J. Goelitz And H. N. Mayrovitz, “The Impact Of
An Emotional Self-Management Skills Course On Psychosocial Functioning And Autonomic Recovery
To Stress in Middle School Children,” Integrative Physiological And Behavioral Science, Cilt 34, No.
4, Pp. 246-268, 1999.
[16] Ö. Kıvrak, F. Ersöz and T. Ersöz, “Karabük İlinde Deprem Farkındalığına Yönelik Deprem
Eğitiminin Öğrenciler Üzerindeki Etkisinin Araştırılması,” 1. Uluslararası Afet Yönetimi Kongresi,
Gümüşhane, 2018.
210
[17] F. Ersöz, T. Ersöz, Ö. Kıvrak and S. Kardeş, “Kuzey Anadolu Fay Hatti Üzerinde Yer Alan
Karabük İli ve Çevresinde Meydana Gelen Depremlerin Veri Madenciliği ve İstatistiksel Yöntemlerle
İncelenmesi,” Uluslararası Sosyal Araştırmalar Dergisi, cilt 9, no. 47, pp. 353-363, 2016.
[18] F. Aydın, “İlköğretim Sekizinci Sınıf Öğrencilerinin 'Deprem' Algılamaları,” Erzincan
Üniversitesi. Fen Edebiyat Fakültesi, Türk Dili ve Edebiyatı Bölümü Dergisi, 2010.
[19] K. Ronan and D. Johnston, “Corraletes of Hazard Education Programs for Youth”, Risk
Analysis, cilt 21, no. 6, 2001.
[20] E. Esgin, Z. Baba, N. Aytaç and E. Turan, “İnternet Tabanlı Kaynakların Doğruluğu ve
Güvenilirliği Hakkındaki Farkındalığın İncelenmesi,” 5th International Computer & Instructional
Technologies Symposium, Fırat University, ELAZIĞ, 2011.
[21] A. Demirci and S. Yıldırım, “İstanbul'da Ortaöğretim Öğrencilerinin Deprem Bilincinin
Değerlendirilmesi,” Milli Eğitim, cilt 1, no. 207, pp. 89-118, 2015.
[22] R. Shaw and M. Kobayashi, “Role of School in Creating Earthquake-Safer Environment,”
Disaster Menagement and Educational Facilities, pp. 1-7, 2001.
[23] J. Rogstadius, M. Vukovic, C. A. Teixeira, V. Kostakos, E. Karapanos and J. A. Laredo,
“CrisisTracker: Crowdsourced Social Media Curation For Disaster Awareness,” IEEE Xplore Digital
Library, cilt 5, no. 4, 2013.
[24] K. Ozeki and T. Ojima, “Current Status of Disaster Preparedness of Pharmacies and
Differences in Disaster Awareness Based on Pharmacy Size." Disaster Medicine and Public Health
Preparedness (2019): 1-5.,” Disaster Medicine and Public Health Preparedness, pp. 1-5, 2019.
[25] V. Clerveaux and B. Spence, “The Communication of Disaster Information And
Knowledge To Children Using Game Technique: The Disaster Awareness Game (DAG),” International
Journal Of Environmental Research, cilt 3, no. 2, pp. 209-222, 2009.
[26] K. Duong, “Disaster Education And Training Of Emergency Nurses in South Australia,”
Australasian Emergency Nursing Journal, cilt 12, no. 3, pp. 86-92, 2009.
[27] D. Şchiopu, “Applying TwoStep Cluster Analysis for Identifying Bank Customers’
Profile,” Universittatii Petrol – Gaze din Ploieşti Ştiinte Economice, cilt LXII, no. 3, pp. 66-75, 2010.
[28] J. Han, M. Kamber and J. Pei, “Data Mining Concepts and Techniques,” Morgan Kaufmann
Publishers Inc, 2012.
211
[29] İ. Ertuğrul, A. Organ and A. Şavlı, “Veri Madenciliği Uygulamasına İlişkin PAÜ
Hastanesinde Hasta Profilinin Belirlenmesi,” Pamukkale Üniversitesi Mühendislik Bilimleri Dergisi,
cilt 19, no. 2, pp. 97-103, 2013.
[30] S. Giray and F. Esin Gülel, “Avrupa Ülkelerinin İntihar Oranlarına Göre Sınıflandırılması,”
Süleyman Demirel Üniversitesi Fen-Edebiyat Fakültesi Sosyal Bilimler Dergisi, cilt 31, pp. 235-247,
2014.
[31] R. Blashfield and M. Aldenferder, “The Literature on Cluster Analysis,” Multivariate
Behavioral Research, cilt 13, pp. 271-295, 1978.
[32] A. Uslu, C. Çetinkaya, E. Özceylan and S. K. İşleyen, “Organize Sanayi Bölgelerinin
Hiyerarşik-K-Ortalamalar Yöntemi ile Analizi,” Türk Sosyal Bilimler Araştırmaları Dergisi, cilt 2, no.
1, pp. 20-37, 2017.
[33] R. Johnson and D. Wichern, Applied Multivariate Statistical Analysis, New Jersey:
Prentice-Hall, 2002.
[34] S. Rundle-Thiele, K. Kubacki, A. Tkaczynski and J. Parkinson, “Using Two-Step Cluster
Analysis to Identify Homogeneous Physical Activity Groups,” Marketing Intelligence & Planning, cilt
33, no. 4, pp. 522-537, 2015.
[35] A. Tkaczynski, “Segmentation Using Two-Step Cluster Analysis,” Segmentation in Social
Marketing, p. 109–125, 2017.
[36] S. Savaş and N. Topaloğlu, “Veri Madenciliği Yöntemi ile GSM Şebekelerinin Performans
Analizi,” Gazi Üniversitesi Mühendislik Mimarlık Fakültesi Dergisi, Cilt 26, No. 4, pp. 741-751, 2011.
[37] Z. Ceylan, S. Gürsev and S. Bbulkan, “İki Aşamalı Kümeleme Analizi ile Bireysel
Emeklilik Sektöründe Müşteri Profilinin Değerlendirilmesi,” Bilişim Teknolojileri Dergisi, cilt 10, no.
4, pp. 475-485, 2017.
[38] R. Alpar, Uygulamalı Çok Değişkenli İstatistiksel Yöntemler, Ankara: Detay Yayıncılık,
2017.
212
Tersanelerde Yalın 6 Sigma ve Uygulanabilirliği
Emre GÜVEN, Ufuk CEBECİ
Department of Industrial Engineering, Istanbul Technical University, TURKEY
Abstract: Teknoloji kavramının insanoğlunun hayatına girmesiyle teknolojinin kullanım
alanları gün geçtikçe artmıştır. Gelişen teknik beceriler ile üretimin odağı değişmiştir. Zamanla
üretimin verimliliği tartışılmaya başlamıştır. Bu tartışmalar sonucunda endüstriyel hayata
kalite, optimizasyon, verim gibi yeni kavramlar girmiştir. Üretim süreçleri üzerine yapılan
araştırmalarda daha kısa sürede, daha az maliyetli ve daha kaliteli üretim için yeni fikirler ortaya
çıkmıştır. İkinci Dünya Savaşı’nın her alanda olduğu gibi üretim yöntemleri üzerinde de etkisi
olmuştur. Artan teknoloji arayışları yeni üretim ve kalite anlayışlarına olan ihtiyacı
doğurmuştur. Bu çalışmada yalın üretim ve altı sigma kavramları incelenmiş, yöntemler
gösterilmiş ve seçilen bir tersanede uygulaması yapılmıştır. Yalın üretim felsefesi, üretim süreci
boyunca oluşan artık ve katma değersiz durumları elimine etme üzerine oturtulmuştur. Altı
sigma metodoloji ise üretim çıktısındaki hataların azaltılmasını ve üretimin standartlaştırarak
hızlanmasını hedeflemektedir. Yapılan uygulama sonucunda teorik olarak süreçlerdeki
kayıpların büyük oranda azaldığı ve firmanın iyileştirme çalışmaları doğrultusunda kazanç
sağlayacağı görülmüştür.
Keywords: Tersane, Gemi İnşa, Üretim, Yalın, 6 Sigma, Kalite
REFERENCES
[1] Arunagiri, P., & Gnanavelbabu, A. (2014). Identification of Major Lean Production Waste
in Automobile Industries Using Weighted Average Method. Procedia Engineering.
[2] Bulut, S. (2012). Beyaz Eşya Yan Sanayi Sektöründe ERP ve Yalın Üretim Olgunluğu
Analizi ve Otomotiv Yan Sanayi ile Kıyaslama. Doktora Tezi, İTÜ Fen Bİlimleri Enstitüsü.
[3] Doğan, S., & Demiral, Ö. (2008). Yalın Yöntemler ve Altı Sigmayı İçeren Bütünleşik Bir
Yaklaşım: Yalın Altı Sigma. Atatürk Üniversitesi İktisadi ve İdari Bilimler Dergisi, 22(1), 343-366.
[4] Eaton, M. (2013). The Lean Practitioner's Handbook.
213
[5] Ergün, A. K. (2003). Altı Sigma Metodolojisi ve Türkiye’deki Uygulamaları (Doctoral
dissertation, Fen Bilimleri Enstitüsü).
[6] Firuzan, A. R., Kuvvetli, Ü., & Gerger, A. (2012). Altı Sigma Metodolojisi ve Otomotiv
Sektöründe Bir Uygulama. Journal of Yasar University, 25(7), 4176-4188.
[7] Gecü, B. (2008). İç Lojistik Sistemlerinin Yalın Üretim Bakış Açısıyla Yeniden
Tasarlanması ve Otomotiv Sektöründe Örnek Bir Uygulama. Yüksek Lisans Tezi, İTÜ Fen Bilimleri
Enstitüsü.
[8] Küçükuysal, K. (2012). Otomotiv Yan Sanayi Sektöründe ERP ve Yalın Üretim Analizi.
Doktora Tezi, Fen Bİlimleri Enstitüsü, İstanbul Teknik Üniversitesi.
[9] Kaşıkçı, M.K., Cebeci, U. (2018). A Manufacturing Maturity Analysis in The Shipbuilding
Industry, ICONASH 2018 İstanbul, Second International Conference on New Approaches in Social
Sciences and Humanities,265-268
[10] Siddh, M. M., Soni, G., Gadikar, G., & Jain, R. (2014). Integrating Lean Six Sigma and
Supply Chain Approach for Quality and Business Pperformance. 2nd International Conference on
Business and Information Management (ICBIM).
[11] Töre, Ö. G. (2000). Yalın Üretim ve Otomotiv Sanayinde Kullanılması. İstanbul Teknik
Üniversitesi, Yüksek Lisans Tezi.
[12] Uçan, K. (2014). Otomotiv Yan Sanayisinde Malzeme Besleme Sisteminin Yalın Üretim
Yaklaşımıyla Yeniden Tasarlanması ve Bir Uygulama. Yüksek Lisans Tezi, Fen Bilimleri Enstitüsü.
[13] Wahab, A. N., Mukhtar, M., & Sulaiman, R. (2013). A Conceptual Model of Lean
Manufacturing Dimensions. Procedia Technology.
214
Smart and Green Supply Chain Applications in Enterprises
Deniz MERDİN, Filiz ERSÖZ
Department of Industrial Engineering, Karabük University, TURKEY
Abstract: Increasing the flexibility and effecient of enterprises from procurement to sales
provides a great competitive advantage for meeting consumer demands. Providing competitive
advantage is possible through the effective implementation of innovative technologies of fourth
industrial revolution in the all stages of supply chain process. In this context, the technologies
related to industry 4.0 were mentioned in the study and the differences between the traditional
supply chain and the digital supply chain were determined. In addition, the industry 4.0
applications in the digital and green supply chain are mentioned and the steps that must be
followed in the process of transition to the digital supply chain are indicated.
Keywords: Industry 4.0, IoT, Dgital Supply Chain, Green Supply
REFERENCES
[1] B. P. Santos, F. Charrua-Santos, and T. M. Lima, “Industry 4.0: an overview,” 2018, no.
July.
[2] C. Cimini, G. Pezzotta, R. Pinto, and Sergio Cavalieri, “Service Orientation İn Holonic and
Multi-Agent Manufacturing,” Stud. Comput. Intell., vol. 640, no. December 2018, 2016.
[3] L. S. Dalenogare, G. B. Benitez, N. F. Ayala, and A. G. Frank, “The Expected Contribution
of Industry 4.0 Technologies For İndustrial Performance,” Int. J. Prod. Econ., vol. 204, no. August, pp.
383–394, 2018.
[4] J. Smit, “Directorate General For Internal Policies Policy Department A: Economic And
Scientific Policy Industry 4.0 Study,” 2016.
[5] S. Schrauf and P. Berttram, “Industry 4.0: How Digitization Makes The Supply Chain More
Efficient, Agile, And Customerfocused,” Strateg. &Technology, p. pg. 1-32, 2016.
215
[6] R. Burke, A. Mussomeli, S. Laaper, M. Hartigan, and B. Sniderman, “The Smart Factory
Responsive, Adaptive, Connected Manufacturing,” 2017.
[7] L. Ardito, A. M. Petruzzelli, U. Panniello, and A. C. Garavelli, “Towards Industry 4.0
Mapping Digital Technologies For Supply Chain Management-Marketing İntegration,” Bus. Process
Manag. J., vol. 29, no. 5, pp. 323–347, 2009.
[8] E. Manavalan and K. Jayakrishna, “A Review of Internet Of Things (Iot) Embedded
Sustainable Supply Chain For Industry 4.0 Requirements,” Comput. Ind. Eng., vol. 127, no. November
2017, pp. 925–953, 2019.
[9] D. Ivanov, A. Dolgui, and B. Sokolov, “The Impact of Digital Technology And Industry 4.0
On The Ripple Effect And Supply Chain Risk Analytics,” Int. J. Prod. Res., vol. 57, no. 3, pp. 829–846,
2019.
[10] J. C. Bendul and H. Blunck, “The Design Space of Production Planning And Control For
İndustry 4.0,” Comput. Ind., vol. 105, pp. 260–272, 2019.
[11] A. Kattepur, A. Mukherjee, and P. Balamuralidhar, “Verification and Timing Analysis of
Industry 4.0 Warehouse Automation Workflows,” IEEE Int. Conf. Emerg. Technol. Fact. Autom.
ETFA, vol. 2018-September, pp. 1297–1304, 2018.
[12] B. Surajit and A. Telukdarie, “Business Logistics Optimization Using Industry 4.0: Current
Status and Opportunities,” IEEE Int. Conf. Ind. Eng. Eng. Manag., vol. 2019-December, pp. 1558–
1562, 2019.
[13] A. Yıldız, E. U. Ergül, C. Gezegin, and H. Dirik, “Akıllı Depolar için PLC Ünitelerinin
Bulanık TOPSIS Yöntemiyle Değerlendirilmesi,” in III. Uluslararasi Mesleki ve Teknik Bilimler
Kongresi, 2018, October, pp. 65–71.
[14] H. Aydemir, “Sanayi 4.0 ve Türkiye Ekonomisi Açısından Etkileri,” Sosyoekonomi, vol.
26, no. 36, pp. 253–261, 2018.
[15] M. H. Calp, E. Bahçekapılı, and M. Berigel, “Endüstri 4.0 Kapsamında Akıllı Fabrikaların
İncelenmesi,” in 5th International Management Information Systems Conference - MISC 2018, 2018,
pp. 16–20.
[16] B. N. Doğru and O. Meçİk, “The Effects Of Industry 4.0 on The Labor Market In Turkey:
Firm Predictions,” Suleyman Demirel Univ. J. Fac. Econ. Adm. Sci., vol. 23, pp. 1581–1606, 2018.
216
[17] M. M. Erdoğdu, “Türkiye’nin Dördüncü Sanayi Devrimi Sürecindeki Konumu,” In Ii.
International Symposium On Economics, Politics And Administration - ISEPA’18, 2018, no. December,
pp. 1–10.
[18] K. Bär, Z. N. L. Herbert-Hansen, and W. Khalid, “Considering Industry 4.0 Aspects İn The
Supply Chain For An SME,” Prod. Eng., vol. 12, no. 6, pp. 747–758, 2018.
[19] D. Ivanov, S. Sethi, A. Dolgui, and B. Sokolov, “A Survey On Control Theory Applications
To Operational Systems, Supply Chain Management, And Industry 4.0,” Annu. Rev. Control, vol. 46,
pp. 134–147, 2018.
[20] B. Wang and J. E. Ha-Brookshire, “Exploration of Digital Competency Requirements
within the Fashion Supply Chain with an Anticipation of Industry 4.0,” Int. J. Fash. Des. Technol. Educ.,
vol. 11, no. 3, pp. 333–342, 2018.
[21] A. Phase and N. Mhetre, “Using IoT in Supply Chain Management,” Int. J. Eng. Tech., vol.
4, no. 2, pp. 2395–1303, 2018.
[22] M. Shafique, A. Rashid, I. Bajwa, R. Kazmi, M. Khurshid, and W. Tahir, “Effect of IoT
Capabilities and Energy Consumption behavior on Green Supply Chain Integration,” Appl. Sci., vol. 8,
no. 12, p. 2481, 2018.
[23] V. L. da Silva, J. L. Kovaleski, and R. N. Pagani, “Technology Transfer in the Supply Chain
Oriented to Industry 4.0: A Literature Review,” Technol. Anal. Strateg. Manag., vol. 31, no. 5, pp. 546–
562, 2018.
[24] S. Luthra and S. K. Mangla, “Evaluating Challenges to Industry 4.0 Initiatives For Supply
Chain Sustainability in Emerging Economies,” Process Saf. Environ. Prot., vol. 117, pp. 168–179, 2018.
[25] M. M. Erdogdu and S. Akar, “Dördüncü Sanayi Devrimi Döneminde Sürücüsüz Otonom
Araçların Potansiyelleri ve 7 Geleceği: Türkiye Örneği,” in Current Debates In Tourism & Development
Studies, December 2017, pp. 275–298.
[26] R. Accorsi, M. Bortolini, G. Baruffaldi, F. Pilati, and E. Ferrari, “Internet-of-things
Paradigm in Food Supply Chains Control and Management,” Procedia Manuf., Vol. 11, June, pp. 889–
895, 2017.
[27] O. Elmacı, “How Is Provided Sustainable Competitive Advantage In Business? As The
Strategic Planning Tool Bsc Model Proposal in Perspective Industry 4.0,” in 3rd International Congress
on Social Sciences, China to Ardiatic, 2018, no. October 2016, pp. 349–370.
217
[28] M. Maslarić, S. Nikoličić, and D. Mirčetić, “Logistics Response to the Industry 4.0: the
Physical Internet,” Open Eng., vol. 6, no. 1, pp. 511–517, 2016.
218
Bitcoin Price Prediction by Using Artificial Neural Networks and Time Series
Derya SAĞ, Ufuk CEBECİ
Department of Industrial Engineering, Istanbul Technical University, TURKEY
Abstract: Bitcoin is the most popular cryptocurrency in the market. Satoshi Nakamoto has
created the Bitcoin in 2009. Bitcoin movements are a widely discussed topic nowadays. On the
other hand, Today machine learning is effectively deployed in an extensive variety of fields
from natural language processing to image processing, from medical applications to activity
recognition. In this research, we collected and used data about the market value of Bitcoin.
Also, Bitcoin is briefly described in this study. Artificial neural networks and time series
techniques used to estimate the market value of Bitcoin. Finally, The paper concludes with
critical considerations of recent developments and some recommendations for future
researches.
Keywords: Machine Learning, Big Data, Data Mining, Bitcoin
Introduction
There are a lot of coin in the cryptocurrency in the market. Bitcoin is still the leader of market.
The first transaction actualize between Satoshi Nakamoto and Hal Finney ın 2009. In this
research, Bitcoin price forecasting was made by using multilayer perceptron and time series.
Sean McNally; Jason Roche ; Simon Caton (2018) are published an article about the prediction
of Bitcoin with using RNN and LSTM. The LSTM algorithm achieves the highest classification
accuracy [1]. Isaac Madan, Shaurya Saluja, Aojia Zhao (2015) are published an article at
Stanford University. Their data set consists of over 25 features relating to the Bitcoin price and
payment network over the course of five years, recorded daily. Using this information they were
able to predict the sign of the daily price change with an accuracy of 98.7% [2]. Alex Greaves,
Benjamin Au (2015) are published an article that performed to analyze the network’s influence
on overall Bitcoin price. In this paper, They investigated the predictive power of blockchain
network-based features on the future price of Bitcoin. they obtained up-down Bitcoin price
219
movement classification accuracy of roughly 55% [3]. Huisu Jang ; Jaewook Lee (2017) are
published an article that An Empirical Study on Modeling and Prediction of Bitcoin Prices With
Bayesian Neural Networks Based on Blockchain Information. They measured the effect of
BNNs by analyzing the time series of Bitcoin process [4].
Ifigeneia Georgoula(2015) is published an article that title is Using Time-Series and
Sentiment Analysis to Detect the Determinants of Bitcoin Prices [5].
Blockchain
Blockchain is a distributed data logging system that provides encrypted transaction tracking. It
is not a database because saved data cannot be changed or deleted again. Blockchain allows us
to keep records that cannot be changed and manipulated. And what makes this technology so
great is that it doesn't need a central authority.
We can summarize this process in 3 main steps.
1. Creation of Transaction
2. Confirmation of Transaction
First, check whether the Bitcoins referenced in the transaction are used before. Secondly, it is
checked whether the signature in the transaction is correct. This is found by putting the sender's
open address, transaction and signature in a function. If this signature returns to true, this
transaction is put into the approved transaction pool. The next step is to add this transaction to
the chain.
3. The addition of the block to the chain.
You can think of what we call a block as a text file. This text file contains the block number,
the proof of work (POW) number, the proof of work number of the previous block, and finally
the approved transactions.
I.
Bitcoin
We can say that Bitcoin is simply a non-state currency behind it. The Bitcoin network consists
of client computers that are actually connected to this system. Every computer involved in the
network communicates with other clients that are close to it, and Peer-to-Peer starts to exchange
220
data in a way that automatically starts downloading all of the operations that have been done in
the network since 2009, so they begin to synchronize with the network. The list of all
transactions made throughout history is open to everyone in a transparent manner. It is even
possible that you can see some of the web sites currently performing live.
II. Weka
In this reserach, We used to Weka for Bitcoin price prediction. WEKA is one of the packages
used in machine language which is one of the important subjects of computer science.
Developed on the JAVA language as open source at Waikato University and distributed under
GPL license. The name comes from here and consists of the initials of the words Waikato
Environment for Knowledge Analysis. We downloaded the deep learning and time series
packages for Bitcoin price prediction.
Multilayer Perceptron
There are tree layer.
Fig 1. Multilayer Perceptron Graph.
221
Long Short Term Memory (LSTM)
III. Loss Functions
Fig 2. Types of Lost Functions.
Activation Functions
The purpose of activation function is to introduce non- linearity into the network. Non-linearity
allow us to approximate arbitrarily complex functions.
The sigmoid (logistic) function is a very common choice for feed-forward NNs that need
to output only positive values. Although its widespread use, the hyperbolic tangent or ReLU
function are generally more convenient. The values of the function are limited to 0 to 1.
One of the most important activation functions is the hyperbolic tangent function. The
values of the function are limited to -1 to 1. Its shape is similar to the sigmoid function. There
are some advantages over sigmoid function. These include derivatives used in the training of
the NN.
Because of the fact that ReLU is a linear, non-saturating function, It has some
advantages over other activations functions. On the contrary of the sigmoid or hyperbolic
tangent activation functions, ReLU does not saturate to -1, 0 or 1.
222
Fig 3. Sigmoid, Hyperbolic Tangent, ReLU and their derivative graphics.
Other Parameters
Stochastic Gradient Descent: Because of the fact that BGD is a slow algorithm, we prefer to
use stochastic gradient descent for faster calculation. Because of the fact that GD method only
taking a single step for one pass over the training set, it could be a very costly method for a
large data set. SGD never converges as BGD does, it is moving around to some close area global
minimum.[31]
J ( ; x ( i ) ; y
(i)
)
(1)
Learning Rate: Small learning rate converges slowly and gets stuck in false local minima.
Large learning rates can converge more quickly but large learning rates overshoot, become
unstable and diverge. This can cause J to increase, rather than decrease monotonically. Instead
of large or small learning, we can use adaptive learning rates. The optimal solution for the
learning speed is to initially keep the learning rate high and then gradually decrease. When the
learning speed is too small at the beginning, it can be attached to the local optimum value,
causing the global optimum value not to be reached at all. The learning rate value is generally
used as the default value of 0.01, which is reduced to 0.001 after a certain epoch.
Momentum: Momentum is a method that aid to accelerate SGD in the relevant direction and
decreased oscillations as can be seen in the figure shown below.
223
Fig 4.
v
t
SGD without momentum
v
t1
SGD with momentum.
J ( )
(2)
parameter get value between from 0 to 1. It generally get 0.9 and it calculates how much of
the previous gradients being into the calculation.
Adam: Adaptive Moment Estimation is another approach that calculates adaptive learning rates
for every parameters.
Application Results
Table 1. Parameters and Their Values For Application.
Parameters
Value
Batch-size
64
Learning rate
0.001
Momentum
0.9
Number of epoch
100
Optimization algorithm
Stochastic Gradient Descent
Weight initialization method
XAIVER
Updater for SGD
Adam
Initial weight distribution
Normal distribution
Multi-layer perceptron algorithm prediction are listed in the table for 2 weeks.
224
Table 2. Prediction And Actual Value Of Bitcoin Price For 2 Weeks.
Prediction of
Bitcoin Value
Actual Bitcoin
Value
04-03-2018
11360.3894
11.512,60
05-03-2018
11196.8307
11.704,10
06-03-2018
10995.2038
11.500,10
07-03-2018
10794.2281
10.929,50
08-03-2018
10754.9553
10.147,40
09-03-2018
10717.0087
9.466,35
10-03-2018
10544.1645
9.531,32
11-03-2018
10458.7118
9.711,89
12-03-2018
10210.3776
9.937,50
13-03-2018
10074.329
9.470,38
14-03-2018
9914.8471
9.355,85
15-03-2018
9911.4377
8.428,35
16-03-2018
9976.0603
8.585,15
17-03-2018
9862.0574
8.346,53
Date
Fig 5. Closing Price of Bitcoin Graph.
225
Conclusion
In this research, we used 1772 daily closing price of Bitcoin between from 27/04/2013 to
03/03/2018. According to results, Multilayer perceptron has better results than the others.
REFERENCES
[1] S., J., & S. (2018, June 07). Predicting the Price of Bitcoin Using Machine Learning.
Retrieved April 12, 2019, from https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8374483
[2] Madan, I., Saluja, S., & Zhao, A. (2014). Automated Bitcoin Trading via Machine Learning
Algorithms.
[3] Greaves, A.& Au,. (2015, December 8). Using the Bitcoin Transaction Graph to Predict the
Price of Bitcoin.
[4] Jang, H., & Lee, J. (2017, December 4). An Empirical Study on Modeling and Prediction of
Bitcoin Prices With Bayesian Neural Networks Based on Blockchain Information. IEEE Xplore, 6,
5427-5437. doi:10.1109/ACCESS.2017.2779181
[5] Georgoula, Ifigeneia; Pournarakis, Demitrios; Bilanakos, Christos; Sotiropoulos, Dionisios
N.; and Giaglis, George M., "Using TimeSeries and Sentiment Analysis to Detect the Determinants of
Bitcoin Prices" (2015). MCIS 2015 Proceedings. 20.
[6] Niu, F., Recht, B., Christopher, R., & Wright, S. J. (2011). Hogwild! : A Lock-Free Approach
to Parallelizing Stochastic Gradient Descent, 1–22.
226
Analysis of Presence of Bank Branches According to Settlement in
Turkey with Data Mining
Süleyman DÜNDAR, Seliha Seçil BAYRAM
Department of Business Administration, Karabük University, TURKEY
Abstract: Banks are one of the most important actors of the financial system. Banks can
perform their services through alternative distribution channels such as branches, ATMs,
internet banking and telephone banking. Banks, which are affected by economic developments
and who are both financial market actors and profit-making enterprises and employing
employment, provide the most efficient services through their branches. The choice of the
banks' location is very important in terms of bank success. One of the factors affecting the
choice of establishment location of banks is the population of the place. The closest selection
criteria are the GDP per capita in the region and the activities of competing banks. It is important
to choose the place of operation of the banks as well as the widespread branch network in which
the customers will work. In the competitive environment, especially corporate customers work
with more than one bank. It is important that the banks where they work for the enterprises that
have active or widespread networks in the field have branches together in the same settlement.
The aim of this study is; a total of 960 settlements with Turkey's 81 provinces, the districts and
towns, bank examination for the presence of the branches of activity and which banks that
operate together on the same settlement "data mining" One of the suitable ones association rules
"Apriori Algorithm" and is to be determined. For this purpose, the "The Banks Association of
Turkey" in the number of audits and the existence of settlements in which they operate
according to their location by the banks operating in Turkey "IBM SPSS Modeler" has been
analyzed with the program. According to the results obtained, the banks operating together in
the same settlement area may determine the bank preferences of the financial managers.
Keywords: Bank Branches, Data Mining, Association Rules, Apriori Algorithm.
227
REFERENCES
[1]
Türkiye
Bankalar
Birliği
https://www.tbb.org.tr/tr/bankacilik/banka-ve-sektor-
bilgileri/banka-bilgileri/subeler/65 (erişim tarihi: 01.02.2019)
[2] Türkiye Kalkınma Bankaları Birliği http://www.tkbb.org.tr/banka-genel-bilgileri (erişim
tarihi: 01.04.2019) [3] BDDK, (14.02.2019), TBS Temel Göstergeleri Raporu Aralık 2018,
[3] https://www.bddk.org.tr/ContentBddk/dokuman/veri_0014_39.pdf
[4] Yetiz, Filiz, “Bankacılığın Doğuşu ve Türk Bankacılık Sistemi”, Niğde Üniversitesi İktisadi
ve İdari Bilimler Fakültesi Dergisi, Sayı: 9(2), Nisan 2016.
[5] M.T. Kartal ve S.K. Depren, “Bankacılıkta Şubeleşme Eğilimini Etkileyen Makroekonomik
Faktörlerin Belirlenmesi Türk Mevduat Bankaları Üzerine Bir Araştırma - Determination of Affecting
Macroeconomic Factors of Branching Trend in Banking,” Turkish Studies International Periodical for
the Languages, Literature and History of Turkish or Turkic Volume 12/24, p. 97-120, Ekim 2017.
[6] N.T. Çınar, Kuruluş Yeri Seçiminde Bulanık TOPSIS Yöntemi ve Bankacılık Sektöründe
Bir Uygulama, KMÜ Sosyal ve Ekonomik Araştırmalar Dergisi, 12(18), 37-45,2010.
[7] A. Başar, Ö. Kabak ve Y. İ. Topçu, “Banka Şubeleri İçin Uygun Yer Seçiminin
Belirlenmesine Yönelik Tabu Arama Yaklaşımı: Bir Türk Bankası Uygulaması - Tabu Search Approach
For Locating Bank Branches: An Application For A Turkish Bank,” Journal Of Industrial Engineering
(Turkish Chamber Of Mechanical Engineers), Vol. 26 Issue 3, P2-22. 21p, Temmuz 2015. [8] F. Ersöz,
“Veri Madenciliği Teknikleri ve Uygulamaları,” 72 Tasarım Dijital Basım ve Yayınevi, Basım sayısı:2,
Sayfa sayısı:341, ISBN:978-605-65464-0-2, 2017.
[9] Y.L. Chen, J.M. Chen ve C.W. Tung, “A Data Mining Approach For Retail Knowledge
Discovery With Consideration of the Effect of Shelf-Space Adjacency on Sales,” Decisions Support
Systems, cilt: 42, sayı: 3, "1503-1520, 2006.
[10] R. Agrawal ve R.Srikant, “Fast Algorithms for Mining Association Rules in Large
Databases,” Proceedings of the 20th International Conference on Very Large Databases (VLDB),
Santiago, 487-499,1994.
[11] J. Han ve M. Kamber, “Data Mining: Concepts and Techniques,” Morgan Kaufmann
Publishers, San Francisco, 2001.
228
Artificial Bee Colony Algorithm for Container Loading Problem
Tuğrul BAYRAKTAR1, Filiz ERSÖZ1, Cemalettin KUBAT2
1
2
Department of Industrial Engineering, Karabük University,
Department of Industrial Engineering, Sakarya University, TURKEY
Abstract: A container is the one of the main components of transportation systems. Allocating
items into limited spaces, is a kind of combinatorial optimization problem and container loading
problems is a branch of knapsack problems, in which a set of items are loaded into capacitated
domains. Heuristic approaches are mostly applied to solve knapsack problems due to the
problem complexity. Artificial Bee Colony Algorithm and Genetic Algorithm are successful
for solving object placement issues. The first one can obtain sufficient results as well as the
second one. In this study, the performance of Artificial Bee Colony Algorithm, which is applied
on CLP rarely, is compared with Genetic Algorithm, which is applied on CLP widely, to see
the capability of proposed ABC algorithm for further studies.
Keywords: Artificial Bee Colony Algorithm, Genetic Algorithm, Container Loading Problem.
REFERENCES
[1] Lee, K.Y., and El-Sharkawi, M.A. “Modern Heuristic Optimization Techniques”, WileyInterscience, 2008.
[2] Che, C. H., Huang, W., Lim, A., and Zhu, W., “The Multiple Container Loading Cost
Minimization Problem.”, European Journal Of Operational Research, vol. 214(3), pp. 501–511, 2011.
[3] Arora, J. S., Introduction to Optimum Design, 3rd ed., 2012, pp. 643-655.
[4] McCall, J. “Genetic Algorithms For Modelling And Optimization”, Journal of
Computational and Applied Mathematics, vol. 184(1), pp. 205-222, 2005.
[5] Goncalves, J.F., and Resende, M.G.C., “A Biased Random Key Genetic Algorithm For 2D
and 3D Bin Packing Problems.”, International Journal Of Production Economics, vol. 145(2), pp. 500510, 2013.
229
[6] Karaboga, D., “An Idea Based On Honey Bee Swarm For Numerical Optimization”, Erciyes
University Engineering Faculty Computer Engineering Department Technical Report-TR06, 2005.
[7] Pham, D.T., Ghanbarzadeh, A., Koc, E., Otri, S., Rahim, S., and Zaidi, M., “The Bees
Algorithm - A Novel Tool for Complex Optimization Problems”, Technical Note. Manufacturing
Engineering Centre, Cardiff University, UK, 2006.
[8] Karaboga, D., and Akay, B., “A Comparative Study Of Artificial Bee Colony Algorithm”,
Applied Mathematics and Computation, vol. 214, pp. 108-132, 2009.
230
Image Size Scaling and Feature Transformation Function Application for Image
Processing in Machine Learning
Ömer PİŞGİN, Ali Osman ÇIBIKDİKEN
Department of Computer Engineering, Necmettin Erbakan University, TURKEY
Abstract: With the increase in computational power and big data, studies on artificial
intelligence are increasing day by day. Especially deep learning applications are seen in almost
all areas of our lives. The most successful results of deep learning architectures are in image
processing. Different architectural approaches are tried to make image processing fast. Due to
the fact that video images consist of large capacity data, it is very important to achieve high
performance in these video images. In this study, size reduction function has been proposed
that can reduce the size of the high-quality and large-capacity file data and produce results with
a high accuracy rate. The results of the proposed method were compared in terms of
performance and speed with different architectures in image processing using CNN
(Convolutional Neural Network) algorithm. In addition, an application that uses the
recommended size reduction function has also been developed using the Python programming
language.
REFERENCES
[1] V. L. Quoc, J. Ngiam, A. Coates, A. Lahiri, B. Prochnow, A. Y. Ng, “On Optimization
Methods for Deep Learning”, Proceedings of the 28th International Conference on Machine Learning
(ICML-11). 2011. 265-272, 2011.
[2] M. Courbariaux, Y. Bengio, J. P. David, “Binary Connect: Training Deep Neural Networks
With Binary Weights During Propagations”, NIPS 28, 2015.
[3] A. Calderón, S. Roa, J. Victorino, “Handwritten Digit Recognition using Convolutional
Neural Networks and Gabor filters”, Proc. Int. Congr. Comput. Intell., 2013.
[4] P. Yadav, N. Yadav, “Handwriting Recognition System-A Review”, Analysis, Vol. 114, pp.
36-40, 2015.
[5] Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner, "Gradient-Based Learning Applied to
Document Recognition", Proceedings of the IEEE, 86(11):2278-2324, 1998.
231
Determination of Weights With Fuzzy AHP in the Job Evaluation Process
Ataberk OLCAY1, Muharrem DÜGENCİ1, Mümtaz İPEK2
1
2
Department of Industrial Engineering, Karabük University
Department of Industrial Engineering, Sakarya University, TURKEY
Abstract: Job evaluation; In order to create input for performance evaluation and wage
management in businesses, the skills of job, responsibility, job conditions etc. and formally and
systematically. Business valuation deals with the importance of the work and the added value
it provides to the business rather than the employee. This study was carried out in the iron and
steel industry. Firstly, two-way comparison matrices were formed by Analytical Hierarchy
Process which is one of the Multi Criteria Decision Making (MCDM) methods for the main
factors and sub-criteria to be used when conducting business evaluation. Then, the paired
comparison matrices formed by Fuzzy Analytic Hierarchy Process were reconstructed with
triangular fuzzy numbers and the weights of main factors and sub-criteria were calculated by
Chang Ground Analysis method.
Keywords: Job Evaluation, Multi Criteria Decision Making, Analytical Hierarchy Process,
Fuzzy
REFERENCES
[1] İ. Özdaban, "İş Değerlendirme ve Personel Değerlendirme Üzerine Bir Bulanık Karar
Modeli," İstanbul Teknik Üniversitesi, Fen Bilimleri Üniversitesi, İstanbul, 2010.
[2] A. Alkan, "İş Değerlendirme Sürecinde Bulanık Analitik Hiyerarşi Prosesi ile Bütünleşik
Bulanık Topsis Uygulaması," Erciyes Üniversitesi, Fen Bilimleri Enstitüsü, Kayseri, 2012.
[3] A. Spyridakos, Y. Siskos, D. Yannacopoulos ve A. Skouris, "Multicriteria Job Evaluation
For Large Organizations," European Journal of Operational Research, no. 130, pp. 375-387, 2001.
[4] M. Dağdeviren, D. Akay ve M. Kurt, "İş Değerlendirme, Faktör Derece Puanlarının
Belirlenmesinde Hedef Programlama Yönteminin Kullanılması," Gazi Üniv. Müh. Mim. Fak., cilt 19,
no. 1, pp. 89-95, 2004.
232
[5] G. Kayhan, "İnsan Kaynakları Performans Değerlendirilmesinde Bulanık AHP/Bulanık
Topsis ile Hibrit Bir Yapının Oluşturulması ve Bir Uygulama," Erciyes Üniversitesi, Fen Bilimleri
Enstitüsü, Kayseri, 2010.
[6] B. Das ve A. G. Diaz, "Factor Selection Guidelines For Job Evaluation: A Computerized
Statistical Procedure," Computers & İndustrial Engineering, no. 40, pp. 259-272, 2001.
[7] S. Krishnamoorthi ve S. K. Mathew, "Business Analytics And Business Value: A
Comparative Case Study," Information & Management, Cilt 55, no. 5, pp. 643-666, 2018.
[8] S. Gupta ve M. Chakraborty, "Job Evaluation in Fuzzy Environment," Fuzzy Sets And
Systems, Cilt 100, No. 1-3, pp. 71-77, 1998.
[9] K. Tümay Ateş, "Çok Kriterli Karar Verme Teknikleri ile Teknoloji Geliştirme Bölgelerinde
Faaliyet Gösteren Firmaların Performans Sıralaması," Çukurova Üniversitesi, Fen Bilimleri Enstitüsü,
Adana, 2018.
[10] B. Erokutan, "Mavi Yakalı Personel Seçiminde Çok Kriterli Karar Verme Yöntemlerinin
Kullanılması ve Bir Uygulama," Bilecik Şeyh Edebali Üniversitesi, Sosyal Bilimler Enstitüsü, Bilecik,
2016.
[11] H. Şimşek, "Analitik Hiyerarşi Süreci ve Bulanık Analitik Hiyerarşi Süreci Yöntemlerinin
İnsan Kaynaklarının Seçiminde Kullanılması: Güvenlik Sektöründe Bir Uygulama," Süleyman Demirel
Üniversitesi, Sosyal Bilimler Enstitüsü, Isparta, 2015.
[12] B. Denizhan, A. Yılmaz Yalçıner ve Ş. Berber, "Analitik Hiyerarşi Proses ve Bulanık
Analitik Hiyerarşi Proses Yöntemleri Kullanılarak Yeşil Tedarikçi Seçimi Uygulaması," Nevşehir Bilim
ve Teknoloji Dergisi, Cilt 6, No. 1, pp. 63-78, 2017.
[13] F. Sofu, "Bulanık Ortamda Çok Kriterli Karar Verme Yöntemi ile Personel Seçimi:
Havacılık Sektöründe Bir Uygulama," İstanbul Ticaret Üniversitesi, Fen Bilimleri Enstitüsü, İstanbul,
2018.
[14] F. Ahmed ve K. Kilic, "Fuzzy Analytic Hierarchy Process: A Performance Analysis Of
Various Algorithms," Fuzzy Sets and Systems, cilt 361, no. 1, pp. 110-128, 2018.
[15] S. Arıkan Kargı ve Z. B. Aydın, "Bulanık AHP Yönteminin Yenilenebilir Enerji
Alternatiflerinin Seçiminde Kullanılması: Bursa Örneği," Akademik Sosyal Araştırmalar Dergisi, cilt 5,
no. 55, pp. 60-74, 2017.
233
Cost Estimation in the Iron and Steel Industry
Gizem KAPANŞAHİN, Filiz ERSÖZ
Department of Industrial Engineering, Karabük University, TURKEY
Abstract: Sacrifice stands for the production of goods and function, constitutes the costs of
enterprises. Cost is also defined as the provision of consumed goods and functions by a
production enterprise. Accuracy of enterprise activity analysis is very important in order to
make appropriate decisions in enterprises. The consistency of the results ensures correct
decision-making; provides right marketing and competitive advantage. Various elements are
effective in the process of product costing. The items on the basis of product are examined one
by one and the analysis is carried out to obtain the unit costs that reflect the reality. The aim of
this study is to investigate the factors affecting the costs and to estimate the cost in the integrated
system, with data mining classifying models in the process of billet production, in an A
enterprise for the Iron and Steel sector. It is targeted to compare obtained estimation results
with the costs presented inside the enterprise.
Keywords: Cost Estimate, Prediction Methods, Artificial Neural Networks
REFERENCES
[1] Özmaden, M. Ş. ‘Çok Katlı Yapılarda Bina Tanımlayıcı Özellikleri Kullanılarak Monte
Carlo Simülasyon Yöntemi ile Maliyet Tahmini,’ Gazi Üniversitesi, Fen Bilimleri Enstitüsü, Ankara,
2010.
[2] ‘Muhasebe Uygulama,’ [Çevrimiçi]. Available: http://www.muhasebeuygulama.com
/maliyet-gider-harcama-nedir.html. [%1 tarihinde erişilmiştir7 Mart 2019].
[3] Çalışkan, A., Uygulamalı Maliyet Muhasebesi, Ankara: Nobel Yayın Dağıtım, 2005.
[4] Bucak, S., ‘Otomotiv Sektöründe Yapay Sinir Ağı Kullanarak Maliyet Tahmini,’ Sakarya
Üniversitesi, Fen Bilimleri Enstitüsü, Sakarya, 2007.
234
[5] Keçe, F., Ömürbek, V. ve Acar, D. ‘Gri Temelli Maliyet Tahmini,’ Süleyman Demirel
Üniversitesi İktisadi ve İdari Bilimler Fakültesi Dergisi, Cilt 21, No. 2, pp. 453-461, 2016.
[6] Gürsoy, A. ‘Yapay Sinir Ağları Yaklaşımıyla Lastik Kalıbı Maliyetlerinin Tahmin
Edilmesi’, Kocaeli Üniversitesi, Fen Bilimleri Enstitüsü, Kocaeli, 2012.
[7] Karataş, E. ‘Yapay Sinir Ağları ile Yazılım Projesi Maliyet Tahmini,’ İstanbul Üniversitesi,
Fen Bilimleri Enstitüsü, İstanbul, 2011.
[8] Acun, Ö. ‘Gri Temelli Maliyet Tahminin Mobilya Üretim Sektöründe Uygulanması,’
Burdur, 2018.
[9] Mohamed, A. A. S., ‘Improving Cost Calculation in The Iron And Steel Industry In Libya
Usıng The Standard Cost Method,’ University of Craiova, Romania, 2013.
[10] Asmar, M. E. ve Hanna, A. S. ‘New Approach to Developing Conceptual Cost Estimates
for Highway Projects,’ American Society of Civil Engineers, 2011.
[11] Enshassi, A. ‘Factors Affecting the Accuracy of Pre-Tender Cost Estimates in the Gaza
Strip,’ Journal of Construction in Developing Countries, Malaysia, 2013.
[12] Lotfia, E. ‘Cost estimation using ANFIS,’ The Engıneerıng Economıst, 2016. 10
[13] Guo, X. ve Liu, S. ‘Grey Self-memory Combined Model for Complex Equipment Cost
Estimation,’ The Journal of Grey System, 2017.
[14] Feldman, P. ve Shtub, A. ‘Model For Cost Estimation in a Finite-Capacity Environment,’
International Journal of Production Research, 2004.
[15] ¨ggen, A. B., ‘Cost Estimates, Cost Overruns, and Project Continuation Decisions,’ The
Accountıng Revıew, 2016.
[16] Malone, P. K., ‘Applying System Readiness Levels to Cost Estimates -A Case Study,’
2016.
[17] Vurşan, H., ‘Beton Çelik Çubuklarında Mukavemet Özelliklerinin Yapay Sinir Ağları ve
Çoklu Regresyon Yöntemleri ile Tahmini,’ Karabük Üniversitesi, Fen Bilimleri Enstitüsü, Karabük,
2017.
[18] Ersöz, F. Veri Madenciliği Teknikleri ve Uygulamaları, Ankara: Dijital Basım Yayınevi,
2016.
235
[19] Takma, Ç. ve Atil, H. ‘Çoklu Doğrusal Regresyon ve Yapay Sinir Ağı Modellerinin
Laktasyon Süt Verimlerine Uyum Yeteneklerinin Karşılaştırılması,’ Kafkas Üniversitesi Veterinerlik
Fakültesi Dergisi, 2012.
[20] Ayyıldız, M. ‘Yazılım Projeleri Ölçüm Sonuçları Veritabanının Oluşturulması ve Yeni
Yazılım Projelerinin Maliyet Tahmininde Kullanımı,’ Yıldız Teknik Üniversitesi, Fen Bilimleri
Enstitüsü, İstanbul, 2007.
[21] Okutkan, C. ‘Borsa İstanbul Şirketlerinin Hisse Senedi Getirilerinin Yapay Sinir Ağları ve
Çoklu Regresyon Yöntemleri Kullanarak Analizi,’ Kocaeli Üniversitesi, Fen Bilimleri Enstitüsü,
Kocaeli, 2014.
[22] Gökalp, E. ‘Bir Konfeksiyon Fabrikası için Maliyet Tahmini ve Üretim Planlamaya
Aktiviteleri için Bir Karar Destek Sistemi’, Orta Doğu Teknik Üniversitesi, Ankara, 2010.
[23] Ökmen, Ö. ‘İnşaat Projelerimim Belirsizlik Altında Faaliyet Şebeke Çizelgelemesi ve
Erken Maliyet Tahmini: Risk Analizine Dayalı Bir Modelleme Yaklaşımı,’ Gaziantep Üniversitesi,
2008.
[24] Yükseltan, E. ‘Türkiye Elektrik Enerjisi Tüketiminin Zaman Serileri ile Analizi,’ Kadir
Has Üniversitesi, Fen Bilimleri Enstitüsü, 2016.
[25] Gülçiçek, Ü. ‘Yapı Parametrelerinin Değişimi ile Yaklaşık Kaba İnşaat Maliyet Tahmini,’
Sakarya Üniversitesi, Fen Bilimleri Enstitüsü, 2011.
[26] Kurt, B. ‘Tahmin Yöntemleri ile Uçak Verilerinin İşlenmesi,’ Erciyes Üniversitesi, Fen
Bilimleri Enstitüsü, Kayseri, 2018.
236
Determination of Socio-Economic Factors Affecting Forest Fires
(A Case Study of Forest Regional Directorate of Antalya)
Ufuk COŞGUN
Forest Faculty, Forest Policy and Management Department, Karabük University, TURKEY
Abstract: In general, it is seen that the fire data is presented as the amount of the area burned
and the number of fires in time. Flammable material loads in forests, behavior patterns and
models of flammable materials according to climatic conditions etc. works are carried out
rapidly. All these studies aim to manage the process after the fire. There is a need to work on
measures to be taken in order to prevent the occurrence of fire. For this purpose; socioeconomic factors that cause fires should be determined in regions where forest fire is common.
The elimination of these factors will minimize the occurrence of forest fires. In Turkey;
considering that 89% of the forest fires are caused by human beings, the importance of socioeconomic studies in these regions is increasing. In the studies to determine the socio-economic
factors that cause forest fires; In some period, some studies such as multiple regression,
correlation, factor analysis etc. were conducted among some socio-economic data determined
according to the conditions of the region and / or fire numbers in the given period. In all these
studies a time section / part was taken into consideration. Materials and Methods; First time in
Turkey with this work; the relationship between the number of forest areas and forest fire
counts and the socio-economic variables determined by considering a certain period of time
was analyzed together with time and space. In the forest fires in the years 1980-1990-2000 in
Antalya Forest Regional Directorate for twelve governmental forest enterprises, the
relationship between the number of forest areas and fire numbers and 25 socio-economic
variables were determined. These data were analyzed by panel data analysis or Time Series
Cross Section Regression (TSCSREG) analysis method.
Variables with no effect in analysis and at the same time the variables / criteria that were
derived from each other were taken into consideration and these criteria were eliminated by
multiple linkage analysis (multiple linear analysis, multiple collinearity analysis) and reduced
237
to 12 variables. Analysis of burned forest areas and selected socio-economic variables; It was
tested by Fuller and Battese Methods in the scope of TSCSREG Analysis.
Conclusion: In the analysis of the amount of burned forest areas and selected socio-economic
variables; A strong relationship between forest fires and selected socio-economic variables
shows that the value of R2 is 90.18%. The value of R2 is 70.19%. It shows that there is a
relationship between the numbers of fires and the socio-economic variables selected.
Keywords- Forest Fires, Panel Data Analysis, Socio-Economic Factors, Multiple Collinearity
Analysis
Introduction
Turkey’s forests are administered by the Ministry of Agriculture and Forestry. The General
Directorate of Forestry is the largest unit, with a remit to protect, develop and manage the forest
areas with 21.3 million hectares area of Turkey. This unit works with regional organizations
via 27 “Regional Directorates of Forest” at the national level. During the period 1937–2010,
86.769 forest fires were recorded in Turkey, destroying an area of 1.617.701 hectares.
The forests of Turkey constitute a part of the Mediterranean Forests, and Turkey’s forest
belt in the Aegean and Mediterranean Regions shows many ecological similarities with
Mediterranean forests [1]. On the other hand, Turkey’s geographical location and physical
structure shows some differences from the other Mediterranean basin countries. It constitutes
a bridge between Europe and Asia and due to the geological structure, many different
ecosystems appear, resulting in high biological richness with more than ten thousand plant taxa
[2]. Not only ecologic, biologic, geologic and physical differences but also socio-economic
conditions in Turkey affecting the forestry studies have also differences from the European
countries
Similarly, with the countries with Mediterranean type ecosystems [3], at the southern and
western parts of Turkey covered by Mediterranean type ecosystems, fire is the one of the most
important issues of forest management [4]. As a result of this importance, big amount of the
budget dedicated to the forestry studies are reserved to the studies on forest fires [5].
238
In the context of that, an effective fire management is required and for this it is needed to
know the components of fire in detail like climate, socio-economic conditions, vegetation etc.
As the studies on fire in Turkey is observed, it is seen that they were mostly based on fire
statistics [6] [7], [8], [9], [10]; [11], [12], [13], [14], [15], fire behavior, vegetation dynamics
[16], [1], [17], [18], [19], [20], [21]. On the other hand, the studies on socio-economic
conditions, which is strictly connected with forest fires especially with fire occurrence, are very
few [22]; [23]. A study on the regional status of fires in the region of the Eastern Black Sea
region in the field of Bilgili [14].
Nevertheless, the assessment of the situation in the world and Turkey is seen that some of
the fire detection data out of the area and number of fires burning [15]. In the case of Antalya
Regional Directorate of Forestry, it is seen that the distribution of forest fires by years and
seasons is examined [24].
Knowledge on socio-economic factors are especially important for fire prevention studies.
Especially, the fact that 89% of forest fires in Turkey are human induced like negligence,
realized intentionally, carelessly, accidentally, indicates the importance of socio-economic
conditions [25].
In this context, it was aimed to define the effects of socio-economic factors on forest fires
in this work. For this goal the Antalya province was chosen as the study area. Antalya province
takes the first place in terms of the coverage of burnt area with the annual coverage of 2633 ha
between the periods of 2000-2009. Additionally, the biggest forest fire with the coverage of
about 15000 ha in the history of Republic of Turkey occurred in Antalya, Serik and Taşağıl
forest enterprises in 2008.
The literature review was mainly focused on gaining the knowledge from the countries that
confront the same problem and these are the other southern European countries such as Spain
Greece and Italy.
Forest fires in the Mediterranean Europe are mostly related to human activities. More than
90% of fires are originated from either deliberate or involuntary causes. Socio-economic
changes occurring in Europe in the last decades (e.g., abandonment of agricultural lands,
depopulation of rural areas, changes in agriculture and forestry policies, etc.) have driven
239
landscape transformations affecting fire risk levels through processes like e.g., increase of
unmanaged lands, dead and live biomass accumulation, new uses of the forest and natural lands.
In this work we analyzed and attempted modelling the influence of socio-economic factors and
their change overtime on forest fire occurrence in the Mediterranean Europe (EU-Med) [26].
Despite Mediterranean countries being strongly sensitive to fire risk, few researches have
focused on long-term fire risk trends, especially related to socioeconomic development. The
present study offers an integrated time-series analysis along a long temporal series (1961–2017)
exploring forest fires in Italy and the relative socioeconomic and demographic Dynamics.
The results of the related study in Italy are as follows “Number of fire events, total burnt
area and average fire size were studied between 1961 and 2017 in Italy with the aim to identify
homogeneous time periods with similar wildfire frequency and severity and correlate them with
the background socioeconomic context. Fire attributes had a diverging behavior over time: the
number of fires was the highest in the 1970s and the early 1980s; total burnt area was relatively
more constant over time with a peak in the 1980s; and, finally, average fire size decreased quite
homogeneously from the peak observed in the 1960s and early 1970s. The number of fires and
average fire size were significantly influenced by the value of the same variable one year
before. Investigating long-term historical outlines of forest fires, a mixed approach based on
time-series statistical analysis, multivariate techniques and regressive models intended to
define changes in fire regimes and socioeconomic development. In fact, the comparative
valuation of the socioeconomic aspects and wildfire trends can reveal a key step to recognizing
mitigation and preventive possibilities. Through a multivariate analysis, a substantial
difference in the socioeconomic profile can emerge by decade, evidencing a (more or less)
rapid socioeconomic development in relation to the evolution of forest fires in Italy [27].
The results of the study in Spain are as follows “The majority of wildfires (95%) in Spain
are caused by human activities. However, much wildfire research has focused on the biological
and physical aspects of wildfire, with comparatively less attention given to the importance of
socio-economic factors. With recent changes in human activity and settlement patterns in many
parts of Spain, potentially contributing to the increases in wildfire occurrence recently
observed, the need to consider human activity in models of wildfire risk for this region are
apparent. A method from Bayesian statistics used; the weights of evidence (WofE) model, to
240
examine the causal factors of wildfires in the south west of the Madrid region for two
differently defined wildfire seasons. The results show that spatial patterns of wildfire ignition
are strongly associated with human access to the natural landscape, with proximity to urban
areas and roads found to be the most important causal factors [28].
The results of the collaboration of the Spanish and Italian researchers are as follows; “Three
6-year time periods were considered (1988-1993, 1998-2003, 2004-2009). Fire data was
extracted from the European Fire database of the European Forest Fire Information System
(EFFIS). The analysis was performed in the most fire-affected area of Europe, the European
Mediterranean region covered by Portugal, Spain, France, Italy and Greece. Fire data were
analyzed according to their main fire cause (accident-negligence and deliberate). Both, fires
and socio-economic variables, which represent anthropogenic factors related to fire activity,
were mapped on a 10 km x10 km grid. Models of forest fire density were derived separately for
each period and per cause category using ordinal regression statistical methods. The best
predictors by period and fire cause category were assessed and differences between time periods
analyzed. The result show that the variable wildland-urban interface (WUI) is related to higher
ratings of fire density. This result was consistent for all periods and for the two type of fire
cause analyzed. The overall fit of the models was 40- 50%.for accident-negligence and 50%
deliberated caused fires. Despite the relatively homogeneous socioeconomic characteristics of
the Mediterranean Europe, differences are found at regional level and by fire cause category”
[26].
Again, a recent research on the forest fires events in Spain is as follows. During the period
1999-2008 an annual average of close to 8,600 forest fires burned about 40,000 ha in Galicia.
Most fires are human caused (99%), approximately 82% are set intentionally and 5% are either
ignited accidentally or through negligence ([29]. However only a limited number of researches
has specifically evaluated how the human presence in this territory increase the risk of fire
ignition ([30]; [31]. This contrasts with the increasing literature on empirical assessments of the
influence of socioeconomic aspects on forest fire risks, using variables such as population
density, land cover changes associated with agriculture abandonment, distance to road or the
density of human settlements [32]; [33].
241
The general situation in Greece about forest fire events is as follows; Especially in Greece,
where the problem of wildfires is going to be analyzed, the statistics of burnt forest area is
alarming. Throughout the period1970-1997, the average burned area had increased from 11,000
ha/year, which was before1970, to 30,000-80,000 ha/year. The figures in the last three years
increased to about 100,000-220,000 ha/year [34]. The customary land use that shaped the
Mediterranean ecosystems through the centuries had contributed to the wildfire hazard
limitation. Grazing of livestock, collection of firewood for domestic uses, resin collection from
pine forests, were traditional practices that keeping down the biomass accumulation.
Furthermore, watching over the neighbouring forests and firefighting were two actions in which
villagers were traditionally engaged. The changes in land use and demography resulted in a
significant increase in fire hazard [35]. The expansion of the cities and all kinds of productive
infrastructure over these ecosystems caused severe deforestation over the years. As it was
mentioned previously, the number and area burned per year have significantly increased in the
last decades due to rural emigration from the mountainous areas and to the change of the socioeconomics conditions [36]. Consequently, fire shave become more frequent, and more
destructive to the ecosystems than used to be in the past. It is worth to mention that in Greece
during the period from 1967-1975 there were an average number of 558 fires per year, which
has reached to 1,841 per year in the last decade. Contrary to that, recreational use of nature and
tourism, have lately brought people near countryside but with different lifestyles and support
urbanization, often increasing fire risks by negligence. These new categories of settlements,
they do not live off the land, using controlled burning to diminish the progressively
uncontrollable fuels [37]. Summer homes spread over coastlines and tourism replaces
traditional industries; political unrest during national elections, are some paradigms of these
developments. Another aspect of justifying the hypothesis that recent fire activity is not
naturally induced are the years 1981, 1985, 1988 where the total burned area reached to large
numbers (table 3). In those years the Greek general elections took place and according to [38]
periods of political unrest in Greece, mainly the periods before the elections, emphasize the rate
of fires [39].
242
Material and Method
In this work, fire numbers and burnt areas of the forest enterprises of Forest Regional Directory
of Antalya between the years of 1980 and 2010, socioeconomic data of 1980’s, 1990’s and
2000’s and fire crimes data of the same periods were analyzed and the correlations between
those variables were observed. Fire numbers and burnt areas data were obtained by Forest Fire
Fighting Department of the Forest Regional Directory of Antalya [5]. Forest crime data were
captured by the forest enterprises chosen as the study areas and fighting department with forest
hazards of Forest Regional Directory of Antalya [40]. Socioeconomic data were obtained by
the Statistic Institution of Turkey [41], Agriculture Directory of Antalya City [42] and the
Agriculture Directory of the towns of Antalya. In all these works, data observation method was
used.
Forest Regional Directory of Antalya consists of thirteen (13) forest enterprises. Each of
these enterprises deals with the forestry affairs of the towns of Antalya. But, only in one town,
Manavgat town, two forest enterprises known as Manavgat and Taşağıl forest enterprises
appears. The second enterprise in Manavgat town was found in 2000. Due to that,
socioeconomic data was only obtained for the town, but forestry related data like forest crimes,
fire numbers and burnt areas were obtained for two forest enterprises.
The correlation of fire data (fire numbers and burnt area) with socioeconomic data and fire
crime data were carried out by using Panel Data Analysis (TSCSREG- time series cross section
regression-TSCSREG) [43], [44].
However, in the previous studies in Turkey carried out by using multi regression,
correlation and factor analysis [22], [23]. In this analysis, in addition to the fire data, total 25
variables including 13 social, 5 economic and 7 forest crime data were used. These additional
variables were analyzed under two groups (Table 1). Some variables were eliminated by using
multiple linear regressions and multiple collinearity analysis since they don’t have any effects
and derived from each other’s (Table 2).
In this way, the socioeconomic variable effective on the burnt areas (Y1) and fire numbers
(Y2) between the years of 1980-2010 were defined. The correlation between these variables
were determined by using Pearson Correlation Analysis.
243
Table 1. Dependent and Independent (Socio Economic Variables/Factors) Variables
Y1
FIRE NUMBER
Y2
BURNT AREA
F1
The Populations Working in Agriculture
F2
The Populations Working in Industry
F3
The Populations Working in Construction
F4
The Populations Working in Service
F5
Activities not adequately defined
F6
%Unemployment Rate
F7
Agrarian Holders Older Than 55
F8
total age dependency ratio
F9
Literate
F10
literate ratio
F11
Illiterate
F12
illiterate ratio
F13
completing primary school
F14
completing primary school ratio
F15
completing junior high school & higher education
F16
completing junior high school & higher education ratio
F17
Population
F18
population density
F19
Illegal cutting
F20
Transport
F21
Illegal keeping
F22
Illegal consumption
F23
Illegal opening and settlement
F24
Occupation
F25
Grazing
244
Table 2. Dependent and Independent (Socio Economic Variables/Factors) Variables
Y1
FIRE NUMBER
Y2
BURNT AREA
F1
The Populations Working in Agriculture
F2
The Populations Working in Industry
F3
The Populations Working in Construction
F4
The Populations Working in Service
F5
%Unemployment Rate
F6
total age dependency ratio
F7
Literate
F8
completing primary school
F9
Population
F10
Illegal cutting
F11
Illegal opening and settlement
F12
Illegal Grazing
The TSCSREG (Time Series Cross Section Regression) procedure analyzes a class of linear
econometric models that commonly arise when time series and cross-sectional data are
combined. The TSCSREG procedure deals with panel data sets that consist of time series
observations on each of several cross-sectional units. Such models can be viewed as two-way
designs with covariates
where N is the number of cross sections, T is the length of the time series for each cross
section, and K is the number of exogenous or independent variables. The performance of any
estimation procedure for the model regression parameters depends on the statistical
characteristics of the error components in the model. The TSCSREG procedure estimates the
regression parameters in the preceding model under several common error structures. The
245
error structures and the corresponding methods the TSCSREG procedure uses to analyze them
are as follows:
One and two-way fixed and random effects models. If the specification is
dependent only on the cross section to which the observation belongs, such a model is referred
to as a model with one-way effects. A specification that depends on both the cross section and
the time series to which the observation belongs is called a model with two-way effects.
Therefore, the specifications for the one-way model are
and the specifications for the two-way model are
where
is a classical error term with zero mean and a homoscedastic covariance
matrix.
Apart from the possible one-way or two-way nature of the effect, the other
dimension of difference between the possible specifications is that of the nature of the crosssectional or time-series effect. The models are referred to as fixed effects models if the effects
are nonrandom and as random effects models otherwise.
first-order autoregressive model with contemporaneous correlation
The Parks method is used to estimate this model. This model assumes a first-
order autoregressive error structure with contemporaneous correlation between cross sections.
The covariance matrix is estimated by a two-stage procedure leading to the estimation of
model regression parameters by GLS.
Mixed variance-component moving average error process
uit= ai+ bt+ eit
246
The Da Silva method is used to estimate this model. The Da Silva method
estimates the regression parameters using a two-step GLS-type estimator.
The TSCSREG procedure analyzes panel data sets that consist of multiple time series
observations on each of several individuals or cross-sectional units. The input data set must
be in time series cross-sectional form. " Working with Time Series Data," for a discussion of
how time series related by a cross-sectional dimension are stored in SAS data sets. The
TSCSREG procedure requires that the time series for each cross section have the same number
of observations and cover the same time range.
Fuller method is most commonly used in analysis Usually you cannot explicitly specify all
the explanatory variables that affect the dependent variable. The omitted or unobservable
variables are summarized in the error disturbances. The TSCSREG procedure used with the
Fuller-Battese method adds the individual and time-specific random effects to the error
disturbances, and the parameters are efficiently estimated using the GLS method. The variance
components model used by the Fuller-Battese method is
The following statements fit this model. Since the Fuller-Battese is the default method, no
options are required.
Results and Discussion
Development of efficient forest fire policies requires an understanding of the underlying
reasons behind forest fire ignitions. Globally, there is a close relationship between forest fires
and human activities, i.e., fires understood as human events due to negligence (e.g.,agricultural
burning escapes), and deliberate actions (e.g., pyromania, revenge, land use change attempts).
Wildfire occurrence even for human-ignited fires has also been shown to be dependent on
biophysical variables [45].
Relations between burning forest areas and socio-economic factors
Analysis of selected socio-economic variables with the amount of forest areas was tested by
TSCSREG analysis and Fuller and Battese Method (Table 3, 4, 5, 6 and 7). There is a strong
247
correlation between the amount of forest areas burned in forest fires and the socio-economic
variables selected and shows that the value of R2 is 90.18% (Table 4). When the relationship
between the amount of forest areas and the socio-economic factors selected; F3 (The
Populations Working in Construction), F4 (The Populations Working in Service), F5
(%Unemployment Rate), F6 (Total Age Dependency Ratio) and F9 (Population) factors appear
to be significant (Table 7).
Table 3. Dependent and Independent Variables of Model Description
Model Description
Estimation Method
Fuller
Number of Cross Sections
12
Time Series Length
3
Table 4. Dependent and Independent Variables of Fit Statistics
Fit Statistics
SSE
143.905 DFE
MSE
6.2568 Root MSE
R-Square
0.9018
23
2.5014
Table 5. Dependent and Independent Variables of Variance Components
Variance Component Estimates
Variance Component for Cross Sections
15.4931
Variance Component for Time Series
1.10147
Variance Component for Error
5.69724
248
Table 6. Hausman Test
Hausman Test for Random Effects
DF
m Value
Pr > m
12
11.66
0.4734
Table 7. Dependent and Independent Variables of Parameter Estimates
Parameter Estimates
Variable
DF Estimate
Standard Error t Value
Pr > |t|
Intercept
1
-33.236
5.3823
-6.18 <.0001
F1
1
0.00075
0.00043
1.74 0.0959
F2
1
0.00864
0.00451
1.92 0.0680
F3
1
0.00992
0.00273
3.64 0.0014
F4
1
-0.0049
0.0014
-3.52 0.0018
F5
1
1.21868
0.3812
3.2 0.0040
D6
1
0.4061
0.0687
5.91 <.0001
F7
1
0.00031
0.0004
0.77 0.4469
F8
1
-0.0013
0.00076
-1.68 0.1066
F9
1
0.00045
5.2E-05
8.75 <.0001
F10
1
0.0229
0.0345
0.66 0.5139
F11
1
-0.0331
0.0596
-0.56 0.5841
F12
1
0.00099
0.0276
0.04 0.9716
Relations Between Socio-economic Factors and Numbers of Fire
249
Analysis of socio-economic variables selected by fire numbers was tested by TSCSREG
analysis and Fuller and Battese Method (Tables 8, 9, 10, 11, and 12). There is a relationship
between the number of fires and the socio-economic variables selected and shows that the value
of R2 is 70.19% (Table 10). When the relationship between the number of fire and selected
socio-economic factors; F2 (The Populations Working in Industry), F5 (% Unemployment
Rate), F9 (Population), F10 (Illegal cutting), F12 (Illegal Grazing) factors (significant) factors
appear to be significant (Table 12).
Table 8. Dependent and Independent Variables of Model Description
Model Description
Estimation Method
Fuller
Number of Cross Sections
12
Time Series Length
3
Table 9. Dependent and Independent Variables of Fit Statistics
Fit Statistics
SSE
MSE
R-Square
154364.9674
DFE
23
6711.5203
Root MSE
81.9239
0.7019
Table 10. Dependent and Independent Variables of Variance Components
Variance Component Estimates
Variance Component for Cross Sections
572.9875
Variance Component for Time Series
448.1269
Variance Component for Error
6598.732
250
Table 11. Hausman Test
Hausman Test for Random Effects
DF
m Value
Pr > m
12
14.08
0.2955
Burning forest areas with quantities; There was a high significant correlation between “The
Populations Working in Construction”, “The Populations Working in Service”,
“%Unemployment Rate”, “Total age dependency ratio” and Population” socio-economic
factors.
There is a high significant correlation between the number of fire and “The Populations
Working in İndustry”,” %Unemployment Rate”, “Population”, “Illegal cutting” and “Illegal
Grazing socio-economic factors. it is necessary to carry out similar studies and analyzes in
other areas characterized by intense forest fires in Turkey. As a result of these analyzes;
Policies should be developed for socio-economic variables of high importance.
Table 12. Dependent and Independent Variables of Parameter Estimates
Parameter Estimates
Variable DF Estimate Standard Error t Value Pr > |t|
Intercept
1 121.5446
128
0.95 0.3524
F1
1 -0.01312
0.0112
-1.17 0.2534
F2
1 0.262899
0.1048
2.51 0.0196
F3
1 0.008581
0.0617
0.14 0.8907
251
F4
1 -0.04681
0.0313
-1.49 0.1489
F5
1 -18.6333
8.7657
-2.13 0.0445
F6
1 -0.56673
1.7407
-0.33 0.7477
F7
1 -0.00106
0.0095
-0.11 0.9122
F8
1 -0.01003
0.0175
-0.57 0.5721
F9
1 0.002732
0.00106
2.57 0.0172
F10
1 2.102254
0.7993
2.63 0.0150
F11
1 -0.57851
1.3064
-0.44 0.6620
F12
1
0.7227
-2.9 0.0080
-2.0978
Conclusion
Rapid deterioration in nature, significant developments in climate change, increase the
importance of forest fires among Mediterranean countries. Due to the unique social and
economic structures of the countries, their evaluations about forest fires are different. When
the forest fires are examined; burning field and in the statistics made based on the number of
fires is observed that there are differences between European countries and Turkey [5]. The
causes of the forest fires in Turkey is 85-90% of human origin [23]. But; more resources and
time are devoted to work after the fire. Since 85-90% of the forest fires are of human origin,
more precautionary measures should be taken into consideration. In respect of measures to be
taken before forest fires; especially the socio-economic factors that cause forest fires are of
great importance. It is important to develop training and awareness-raising activities for civil
society organizations, other public institutions and institutions and institutional infrastructures
in terms of sensitivity to forest fires.
Turkey is the most vulnerable region in terms of forest fires in Antalya Regional Forestry
Directorate were examined in the study. Selected socio-economic factors were evaluated
statistically by taking into consideration the time sections. In order to determine the socio252
economic factors causing forest fires, panel data analysis was first evaluated in this study. In
terms of forest fire numbers and burning areas, two different points of view were analyzed.
Examining the relationship between burning forest areas and socio-economic factors;
F3 (The Populations Working in Construction),
F4 (The Populations Working in Service),
F5 (%Unemployment Rate),
F6 (Total Age Dependency Ratio) ve
F9 (Population)
factors were found to be statistically significant differences between the burning forest areas.
F3, F4 and F5 factors are completely related to employment. Again, F9 factor is a factor related
to the population in rural areas. Factors such as the increase in the population and the decrease
of employment are the leading factors leading to the emergence of economic problems.
Employment will increase if employment decreases, the unemployment rate (F5)is
significantly as a factor. Decreasing the share of the population in the employment in the region
will lead to different ways of benefiting from other fields. As a result of increasing burning
forest areas; a new additional employment area for the removal of damaged forest products in
these areas. Therefore, a statistical relationship between these factors and burning forest areas
is quite consistent.
Examining the relationship between the numbers of fire and socio-economic factors
F2 (The Populations Working in Industry),
F5 (% Unemployment Rate),
F9 (Population),
F10 (Illegal cutting),
F12 (Illegal Grazing)
253
There was a statistically significant difference between these factors and fire numbers.
Among the socio-economic factors, F2, F5 and F9 factors (which are related to employment
and population) have been statistically significant factor on fire numbers. Another group of
factors is the illegal utilization of forests (illegal grazing and illegal cutting).98% of forest area
in Turkey is a State Forest. Therefore, there are serious legal regulations against illegal
exploitation from forests. Unlawful beneficiaries from forest areas can develop negative
behaviors towards forest areas in response to criminal practices, such as deliberately burning
forests. In the forestry work, those who are employed are unemployed or the burning of forests
as a result of the punishment of illegal beneficiaries is the realities encountered for the socioeconomic conditions of our country. There are two factors that are found as burning forest areas
and fire numbers, which are common and have statistically significant differences on both
variables. These; F5 (% Unemployment Rate) and F9 (Population). Population growth and
unemployment should be seen as the main factors. As a result of the analysis; The socioeconomic factors that occur in two main groups, especially in terms of fire numbers, are the
factors that should be examined in more detail on the basis of forest villages in the region.
REFERENCES
[1] Kavgacı, A., Carni, A., Başaran, S., Başaran, M.A., Kosir, P., Marinsek, A., Silc, U., 010.
Long-Term Post-Fire Succesion of Pinus Brutia Forest in The East Mediterranean. International Journal
of Wildland Fire (basım aşamasında).
[2] Güner, A. Ozhatay, N. Ekim, T. and H.C. Baser. 2000 (eds). Flora of Turkey and the East
Aegean Islands, Supplement 2. Edinburg University Press.
[3] Anonym, 2010-2013; FUME Project. 7th Framework Programme of European Union.
[4] Anonym, 2010; YARDOP Tamini No: 6665, Çevre ve Orman Bakanlığı, Orman Genel
Müdürlüğü, Sivil kültür Daire Başkanlığı, Ankara.
[5] Anonym, 2012; Orman Yangınları İle Mücadele 2012 Yılı Eylem Planı, Orman ve Su İşleri
Bakanlığı, Orman Genel Müdürlüğü, Ankara.
[6] Baş, R., 1965; Türkiye’de Orman Yangın Problemleri ve Bazı Klimatik Faktörlerin
Ynagınlara Etkileri Üzerine Araştırmalar, OGM, Yayın No: 421/20, Ankara.
254
[7] Çanakcıoğlu, H., 1985; Orman Koruma, İ.Ü. Orman Fakültesi, İ.Ü. Yayın No: 3315, Orm.
Fak. Yayın No: 376, İstanbul.
[8] Çanakcıoğlu, H., 1988; Orman Yangın İstatistikleri, Türkiye Ormanlarını Yangından Koruma
Semineri, Tarım Orman ve Köyişleri Bakanlığı, Orman Genel Müdürlüğü, Orman Koruma ve Yangınla
Mücadele Daire Başkanlığı, Yayın No: 29, Seri No: 672, Ankara.
[9] Küçükosmanoğlu, A., 1987; Türkiye Ormanlarında Çıkan Yangınların Sınıflandırılması ile
Büyük Yangınların Çıkma ve Gelişme Nedenleri, OGM Yayın No: 662, Seri No: 28, Ankara.
[10] Torlakçık, Ş., 1988; İzmir Orman Bölge Müdürlüğü’nde Orman Yangınları ve Yangından
Korunma, Türkiye Ormanlarını Yangından Koruma Semineri, Tarım Orman ve Köyişleri Bakanlığı,
Orman Genel Müdürlüğü, Orman Koruma ve Yangınla Mücadele Daire Başkanlığı, Yayın No: 29, Seri
No: 672, Ankara.
[11] Kayatürk, D., 1988; Çanakkale Orman Bölge Müdürlüğü’nde Orman Yangınları ve
Yangından Korunma, Türkiye Ormanlarını Yangından Koruma Semineri, Tarım Orman ve Köyişleri
Bakanlığı, Orman Genel Müdürlüğü, Orman Koruma ve Yangınla Mücadele Daire Başkanlığı, Yayın
No: 29, Seri No: 672, Ankara.
[12] Toklu, Y., 1988; Antalya Orman Bölge Müdürlüğü’nde Orman Yangınları ve Yangından
Korunma, Türkiye Ormanlarını Yangından Koruma Semineri, Tarım Orman ve Köyişleri Bakanlığı,
Orman Genel Müdürlüğü, Orman Koruma ve Yangınla Mücadele Daire Başkanlığı, Yayın No: 29, Seri
No: 672, Ankara.
[13] Akaydın, S., 1988; Muğla Orman Bölge Müdürlüğü’nde Orman Yangınları ve Yangından
Korunma, Türkiye Ormanlarını Yangından Koruma Semineri, Tarım Orman ve Köyişleri Bakanlığı,
Orman Genel Müdürlüğü, Orman Koruma ve Yangınla Mücadele Daire Başkanlığı, Yayın No: 29, Seri
No: 672, Ankara.
[14] Bilgili, E., Dinç Durmaz, B., Baysal, İ., Sağlam, B., Küçük, Ö., 2010; Doğu Karadeniz
Ormanlarında Orman Yangınları, III. Ulusal Karadeniz Ormancılık Kongresi, Cilt: III, Sayfa: 12801290, Artvin.
[15] Ertuğrul, M., 2005; Orman Yangınlarının Dünyadaki ve Türkiye’deki Durumu, ZKÜ, Bartın
Orman Fakültesi Dergisi, Bartın.
[16] Kavgacı, A., Silc, U., Carni, A., Başaran, S., Başaran, M.A., Kosir, P., Marinsek, A., Zelnik
I., 2009. Serik ve Manavgat Bölgelerindeki Kızılçam Ormanlarında Yangın Sonrasında Vejetasyon
255
Yapısında Meydana Gelen Değişim. 1. Orman Yangınlarıyla Mücadele Sempozyumu Tebliğleri, 7-10
Ocak 2009, Antalya, Orman Genel Müdürlğü Yayını, s. 295-302.
[17] Tavşanoğlu, Ç., 2010. Seed Production and Fruit Parasitism in Cistus Salviifolius L.
(Cistaceae) Along A Post-Fire Successional Gradient. J. An. Vet. Adv. 9(7): 1120-1127.
[18] Tavşanoğlu, Ç., Gürkan, B., 2004. Akdeniz Havzasında Bitkilerin Kuraklık ve Yangına
Uyumları. Ot Sistematik Botanik Dergisi, 11(1):119-132.
[19] Tavşanoğlu, Ç., Gürkan, B., 2005. Post-Fire Dynamics of Cistus Spp. in a Pinus brutia
forest. Turkish J. Bot., 29:337-343.
[20] Tavşanoğlu, Ç., Gürkan, B., 2009. Post-Fire Regeneration of a Pinus Brutia (Pinaceae)
Forest In Marmaris National Park, Turkey. Int. J. Bot., 5:107-111.
[21] Tavşanoğlu, Ç., Kaynaş, B.Y., Gürkan, B., 2002. Plant Species Diversity in A Post-Fire
Successional Gradient in Marmaris National Park, Turkey. In: Viegas, X.V. (Ed.) Proceedings of the
IV. International Conference on Forest Fire Research – 2002 Wildland Fire Safety Summit, 18-23
November, Luso, Coimbra, Portugal.
[22] Öktik, N., 2001; Orman Yangınlarının Sosyo-Ekonomik ve Kültürel Nedenleri, Muğla
Üniversitesi Yayın No: 16, Muğla.
[23] Coşgun, U., Yolcu, H., İ., Tolunay, A., Orhan, K., H., 2010; Antalya Orman Bölge
Müdürlüğünde Orman Yangınlarına Neden Olan Sosyo-Ekonomik Faktörlerin Belirlenmesi, Batı
Akdeniz Ormancılık Araştırma Müdürlüğü, Teknik Bülten No: 40, Antalya.
[24] Coşgun, U., Kavgacı, A., Güngöroğlu, C., 2012; Türkiye’deki Orman Yangın Rejim
Verilerinin Değerlendirilmesi Antalya Orman Bölge Müdürlüğü Örneği, Batı Akdeniz Ormancılık
Araştırma Müdürlüğü Dergisi, Antalya.
[25] Coşgun, U., González-Cabán, A., 2019; Factors Explaining Forest Fires in the Serik and
Taşağıl Forest Provinces (SW Anatolia), Proceedings of Fifth International Symposium on Fire
Economics, Planning and Policy: Ecosystem Services and Wildfirees, USDA, Genral Technical Report
PNW-GTR-’61 (English). February, pages; 145,165.
[26] Vilar L., Camia A., Ayanz, J., S., M., 2014; Modeling Socio-Economic Drivers of Forest
Fires In The Mediterranean Europe, Chapter 7, Advances in Forest Fire Research, Coimbra University
press.
256
[27] Carlucci, M., Zambon, I., Colantoni, A., Salvati, L., 2019; Socioeconomic Development,
Demographic Dynamics And Forest Fires in Italy, 1961–2017: A Time-Series Analysis, Sustainable
11(5), 1305 Case Report.
[28] Calcerrada R., R., Novillo, C., J., Millington J., D., A., Jimenez, I., G., 2008 GIS Analysis
of Spatial Patterns of Human-Caused Wildfire Ignition Risk In The SW Of Madrid (Central Spain),
Landscape Ecol., DOI. 10.1007/s 10980-008-9190-2, Spain.
[29] Chas-Amil M.L., Touza J., Prestemon J.P., 2010. Spatial Distribution of Human-Caused
Forest Fires in Galicia (NW Spain). In G. Perona and C. A. Brebbia (eds.) Modelling, Monitoring and
Management of Forest Fires. WIT Press, 247-258.
[30] Martínez, J., C. Vega-Garcia, E. Chuvieco. 2009. Human-Caused Wildfire Risk Rating for
Prevention Planning in Spain. Journal of Environmental Management, 90, 1241-1252.
[31] Padilla, M. and Vega-Garcia, C., 2011. On the Comparative Importance of Fire Danger
Rating Indices and Their Integration with Spatial and Temporal Variables For Predicting Daily
Human-Caused Fire Occurrences in Spain. International Journal of Wildland Fire, 20(1), 46-58.
[32] Moreira, F., Viedma, O., Arianoutsou, M., Curt, T., Koutsias, N., Rigolot, E., Barbati, A.,
Corona, P., Vaz, P., Xanthopoulos, G., Mouillot, F. and Bilgili, E., 2011. Landscape Wildfire
Interactions in Southern Europe: Implications for Landscape Management. Journal of Environmental
Management, 10, 2389-402.
[33] Narayanaraj, G. and Wimberly M.C., 2012. Influences of Forest Roads on The
Spatialpatterns Of Human- And Lightning-Caused Wilfire Ignitions. Applied Geography, 32, 878-888.
[34] Efthymiou, P., 2000. Mediterranean Forest Fires 2000-A Terrible and Complex Situation.
EFI News No. 2. Volume 8. Finland.
[35] Xanthopoulos, G., 1995. Lectures Notes. Seminar on Forest Fire Prevention and
Security.Educational Society. Athens.
[36] Papanastasis, V.P., 1999. Land Degradation Caused by Overgrazing and Wildfires and
Management Strategies to Prevent and Mitigate Their Effects. Faculty of Forestry and
NaturalEnvironment, Aristotle University. Thessalonica. Greece.
[37] Pyne, S. 1997. World Fire. The culture of fire on Earth. Weyerhaeuser Environmental
Books.University of Washington Press. Seattle and London.
257
[38] Papastavrou, A., 1992. Social, Economic and Cultural Aspects of Forest and Wild land Fires
inGreece. Seminar on Forest Fire Prevention, Land Use and People. Joint Committee on Forest
Technology, Management and Training.
[39] Maheras G., 2002 Forests fires in Greece. The Analysis of The Phenomenon Affecting Both
Natural Andhuman Environment. The Role of Sustainable Development in Controlling Fire Effects. M
Sc. Thesis, LUMES.
[40] Anonym, 2014; Antalya Orman Bölge Müdürlüğü, Koruma Şube Müdürlüğü, 1978-2014
Orman Yangınları Defteri, 2014, Antalya.
[41] Anonym, 2013/a; Türkiye İstatistik Kurumu, İl ve İlçeler Düzeyinde Nüfus ve İstihdam
verileri, Ankara.
[42] Anonym, 2013/b; Tarım İl Müdürlüğü Tarımsal Alan Verileri, Antalya.
[43] Qi, Y., Smith, B.L., and J. Guo 2007. Freeway Accident Likelihood Prediction Using a Panel
Data Analysis Approach, Journal of Transportation Engineering, March 2007, 149-156.
[44] Meserve, S.A., Pemstein, D., and W.T. Bernhard 2009. Political Ambition and Legislative
Behavior in the European Parliament, The Journal of Politics, 71, 3, 1015–1032.
[45] Chas-Amil M., J., Touza J., Prestemon J., P., McClean, C., J.,2012 Natural and Social
Factors Influencing Forest Fire Occurrence At A Local Spatial Scale, Modeling Fire Behavior And Risk
(Project09SEC011201PR).
258
Group Acceptance Sampling Plans Based on Time Truncated Life Tests For Compound
Weibull-Exponential Distribution
Canan HAMURKAROĞLU1, Ayten YİĞİTER2
1
Department of Actuary and Risk Management, Karabük University
2
Department of Statistics, Hacettepe University, TURKEY
Abstract: Acceptance sampling plans are a decision-making process on the basis of a randomly
selected sampling from a party, where it is not possible to completely scan the products for
reasons such as time and cost being limited or the formation of damaged products during the
inspection. For some products, the life span (time from beginning to failure) may be an
important quality characteristic. In this case, the quality control adequacy of the products can
be checked with an acceptance sampling plan based on the truncated life test with a censored
scheme for the lifetime of the products. Acceptance sampling plans based on life test of product
life in industry are called reliability plans. In this study, group acceptance sampling plans based
on life tests were studied under the type-I censored scheme for the compound WeibullExponential distribution. Optimum sample size, optimum number of groups and acceptance
number were obtained.
Keywords: Acceptance Sampling Plan, Truncated Survival Test, Producer Risk, Consumer
Risk, Compound Weibull Exponential Distribution
259
Stochastic Approach to a Buffer Stock Problem
Zülfiye HANALİOGLU1, Tahir HANALİOGLU2
1
Department of Actuary and Risk Management, Karabük University,
2
Department of Industrial Engineering, TOBB University, TURKEY
Abstract: In this study, a buffer stock between two machines which are working at the same
speed, is considered. It is assumed that the stock level between two machines alters in the
interval 𝟎, 𝟐𝒂 . In the case that only the first machine is broken, the stock level can decrease
to zero if repairing time of the first machine is extended. This causes the second machine is
required to be halted. After fixing the first machine, the system begins to work again. If the
second machine is broken and the first machine is working, then the process proceeds until the
stock level reaches the maximum level 𝟐𝒂 and then the first machine will be halt compulsory.
In order to re-work the system, the repairing of the second machine must be completed. Under
these assumptions, the buffer stock level will be stochastically fluctuated in the interval 𝟎, 𝟐𝒂
. Thus, the buffer stock between two machines can be expressed by a stochastic process 𝒀 𝒕 and
it is observed that this process is a random walk with two barriers. Some problems of queuing
theory, stock control, reliability, insurance models and risk management can be expressed by
random walk and its modifications. In literature, there are several considerable studies (e.g.,
Aliyev and Khaniyev, Feller, Janseen and Leeuwarden, etc.). In this study, the stationary
characteristics of the random walk process 𝒀 𝒕 which represent the buffer stock level, are
investigated. Especially, for all moments of the ergodic distribution of the process 𝒀 𝒕, the exact
expressions are obtained under the assumption that the random walk 𝒀 𝒕 generated by bilateral
exponential distributed summands. Moreover, the exact and approximate expressions for
variance, standard deviation, coefficient of variation, skewness and kurtosis of the process, are
obtained. It is also observed that the ergodic distribution of the standardized stochastic process
𝑾 𝒕 = 𝒀 𝒕 /𝒂 weakly converges to a triangular distribution in the interval 𝟎, 𝟐.
Keywords: Buffer Stock, Random Walk, Ergodic Distribution, Stationary Characteristics.
260
REFERENCES
[1] Aliyev R. T. and Khaniyev T. A., 2017, “On the Limiting Behavior of the Characteristic
Function of the Ergodic Distribution of the SemiMarkov Walk with Two Boundaries” Mathematical
Notes, 102 (4), 444-454.
[2] Borovkov A.A.,1984, “Asymptotic Methods in Queuing Theory”, J. Wiley, New York.
[3] Chang J.T. and Peres Y., 1997, “Ladder Heights, Gaussian Random Walks And The
Riemann zeta function”, Annals of Probability, 25, 787–802.
[4] Feller W., 1971, An Introduction to Probability Theory and Its Applications II, John Wiley.
[5] Gihman I.I. and Skorohod A.V., 1975, Theory of Stochastic Processes II, Springer –Verlag.
[6] Janssen A.J.E.M. and Leeuwarden J.S.H., 2007, “On Lerch’s Transcendent And The
Gaussian Random Walk”, Annals Of Applied Probability, 17, 421–439.
[7] Khaniev T. A., Unver I. and Maden S., 2001, “On The Semi – Markovian Random Walk
With Two Reflecting Barriers”, Stochastic Analysis And Applications, 19(5), 799–819.
[8] Khaniev T. A. and Kucuk Z., 2004, “Asymptotic Expansions For The Moments of The
Gaussian Random Walk With Two Barriers”, Statistics And Probability Letters, 69(1), 91 – 103.
[9] Lotov V.I., 1996, “Some Boundary Crossing Problems For Gaussian Random Walks”, The
Annals of Probability, 24(4), 2154–2171.
[10] Spitzer F., 1964, Principles of Random Walk, Princeton, N. J., D. Van Nostrand.
261
Prediction of Air Permeability of Denim Fabrics Using Articifial Neural Networks
Esra AKGÜL, Emel AYDOĞAN, Yılmaz DELİCE
Department of Industrial Engineering, Erciyes University, TURKEY
Abstract: Denim is a popular fabric among all of the age groups because of its good usage
performance and ability to provide convenience in adapting to changing trends in fashion. Apart
from the fashion and general performance properties, thermo-physiological comfort properties
such as air permeability are important for denim users. Fabric comfort depends on lots of factor
such as fabric structure and the types of fibers. Air permeability is a one of the comfort
properties of fabric is affected by many parameters of the fabric. A determination of the
relationship between the fabric parameters and the air permeability is highly complex and
difficult. For this reason, Artificial Neural Network model which has effective performance in
very complex problems was used. In the present study, an artificial neural network has been
used to predict air permeability amongst different denim production parameters. Finally, by
comparison with the experimental results, the efficacy of the proposed model was verified.
Keywords: Artificial Neural Network, Denim Fabric, Optimization.
REFERENCES
[1] Milenkovic L, Skundric P, Sokolovic R & Nikolic T, The Scientific Journal Facta
Universitatis, 1(4), 1999, pp. 101. [2] Oğlakcioğlu, N., & Marmarali, A., “Thermal comfort properties
of some knitted structures”, Fibres & Textiles in Eastern Europe, 15(5-6), 2007, pp. 64-65.
[3] Saville B.P., Physical testing of textiles, The Textile Institute, Woodhead publishing limited,
Cambridge-England, 2003. [4] Çay, A., Vassiliadis, S., Rangoussi, M., & Tarakçıoğlu, I. “Prediction of
the air permeability of woven fabrics using neural networks”, International Journal of Clothing Science
and Technology, 19(1), 2007, pp. 18-35.
[5] TS 391 EN ISO 9237, Tekstil-Kumaşlarda Hava Geçirgenliğinin Tayini, Türk Standartları
Enstitüsü, Ankara, 1999.
262
Numerical Investigation of Cutting Forces in Turning of C23000 Brass Alloy
Mehmet Erdi KORKMAZ
Department of Mechanical Engineering, Karabük University, TURKEY
Abstract: Brass alloys are characterized by excellent workability, high thermal and electric
conductivity, corrosion resistance as well as exceptional antibacterial properties and are
therefore widely used in various industries, such as electric and electronics, automotive and
sanitary industry. However, power consumption should be eliminated for cleaner production in
terms of sustainable machining. Therefore, this study aims modelling of cutting forces in hard
turning of C23000 brass based on finite element method. The cutting parameters are chosen as
cutting speed, depth of cut and feed rate with three levels. The average of 4.66% difference is
achieved between experimental and simulated feed forces while 4.39% difference for main
cutting forces. The finite element modelling of cutting forces is quite compatible with the
experimental results and it can be performed by high accuracy without excessive machining
experiments of high machinability materials.
Keywords: Brass Alloys, Finite Element Method, Main Cutting Force, Feed Force
Introduction
Brass alloys are characterized by excellent workability, high thermal and electric conductivity,
corrosion resistance as well as exceptional antibacterial properties and are therefore widely used
in various industries, such as electric and electronics, automotive and sanitary industry. To
enhance their machinability, lead is commonly added to brass alloys, leading to excellent chip
breakage, low tool wear, and high applicable cutting parameters. Main applications are in
electric and electronics, automotive, and sanitary industry. Since the amount of cutting
operations when manufacturing brass components is high, different alloying elements
enhancing the machinability are usually added to brass [1].
263
The most essential element in this context is lead (Pb), improving the machinability
referring to chip breakage, tool wear, cutting forces, and applicable range of cutting parameters.
The machinability of these low-leaded brasses is significantly worse compared to leaded freecutting brass. Depending on their chemical composition and microstructure, different
machinability problems arise.
The cutting forces are a result of extreme conditions at the toolworkpiece interface. The
interaction can be directly related to the tool wear and in worst cases to the failure of the tool.
Consequently, the tool wear and cutting forces are related to each other. Thus, it is necessary to
carry out the optimization of cutting process to evaluate the optimal values of cutting parameters
to determine the performance and useful life of the cutting tool [2].
Surinder et al. [3] investigated the cutting forces (tangential and feed force) in turning
of unidirectional glass fiber reinforced plastics (UD-GFRP). The process parameters of cutting
tool (nose radius, rake angle, cutting speed, feed rate, depth of cut and cutting environment)
were investigated using Taguchi robust design methodology. The relative significance of
parameters was studied using ANOVA. The tangential force was found to decrease with
decrease in tool nose radius, feed rate and depth of cut and increase with the cutting speed.
Cascona et al. [4] developed mechanistic model for prediction of cutting forces in turning of
non-axis-symmetric parts. This study presents a mechanistic model for predicting the
orthogonal turning forces (in 3 directions), torque and power consumption along the machining
path of non-axis-symmetric parts. Dorlin et al. [5] studied the geometrical modeling of
toolworkpiece interaction and its effects on the cutting forces during turning. The analysis
focused on convex contact radius between the machined part and the tool. The experiments
were based on cylindrical and face turning of Ti6Al4V titanium alloy. It was observed that the
contact radius had significant effect on the cutting forces and the cutting forces increase with
the increase in the radius. Xie et al. [6] studied cutting force and cutting temperature during the
turning of titanium alloy using micro-grooved tool under dry conditions. The objective of the
study was to estimate the influence of shape and size of micro groove on the temperature and
force in dry turning. The micro-grooved tool decreases cutting temperature by 103 _C, while
as the shear angle increases with decreasing micro-groove depth. Philip et al. [7] studied the
effects of cutting speed and feed rate on tool wear, surface roughness and cutting force on
264
nitrogen alloyed duplex stainless steel in a dry turning process, using Taguchi method. The
results revealed that the feed had the most significant influence on the cutting forces. The
cutting speed was found to be the most significant parameter affecting the tool wear. Shear
force, ploughing force and particle fracture force were considered by Sikder et al. [8] to estimate
the cutting force during the machining of metal matrix composites (MMCs). The chip formation
force, ploughing force and fracture force were obtained by Johnson-Cook model, slip line filed
theory and Griffith’s theory respectively. The results showed good agreement between the
predicted and experimental values of the cutting forces.
Literature researches showed that there is no study performed on finite element
modelling on power consumption for C23000 brass alloy. Therefore, in this work, FE modelling
was performed to determine the cutting forces in turning of the material based on cutting
parameters which are cutting speed, feed rate and depth of cut.
Material and Method
The aim of this study was to investigate the effects of cutting parameters on cutting forces in
turning of C23000 brass with HSS tools by using finite element method. For this purpose, the
finite element analysis which are based on experiments by Hanief et al. [2] were performed
with Third Wave Advantedge software with 2D orthogonal turning instead of 3D turning due
to low calculation time [9].
Work Piece Material
The workpiece has hardening of 120 Bhn by means of heat treatment by quenching in a vacuum
atmosphere. The Johnson–Cook model [10] widely-used material model for machining
simulations is given in Eq. 1. This material model is particularly suited to model high strain rate
deformation of metals. It is generally used in adiabatic transient dynamic analysis. The
hardening is a particular type of isotropic hardening in which the yield stress σ0 is assumed as
[11]:
𝜎 = (𝐴 + 𝐵(𝜀 ) ) 1 + 𝐶 𝑙𝑜𝑔
̇
̇
(1)
1−
265
In equation (1), material parameters obtained from mechanical tests that are 𝐴, 𝐵, 𝐶, 𝑛
and 𝑚 are yield stress below room temperature, strain hardening, strain rate constant, strain
hardening constant and thermal softening constant, respectively. The other parameters 𝜀 , 𝜀̇ ,
𝜀̇ , 𝑇 , 𝑇 and 𝑇 are equivalent plastic strain, plastic strain rate, reference strain rate, room
temperature, melting temperature and reference temperature, respectively. Also, 𝜀̇ and 𝐶 are
usually measured at or below the reference temperature. The Johnson-Cook parameters and
other material parameters for AISI 52100 steel was given in Table 1 and Table 2, respectively
from finite element software.
Tablo 1. Johnson–Cook parameters for the C23000 brass alloy material
A
232
B(MPa)
C
n
m
931
0.0682 0.93 0.4682
Tr
27
Tm 𝛆̇ 𝟎
1083 1
Tablo 2. The other material parameters for the C23000 brass alloy material
Density
Poisson
(g/cm3)
ratio
8.4
0.3
Young's
Thermal
Specific
Thermal
Modulus
Conductivity
Heat
Expansion
(GPa)
(W/m.K)
(J/kg°C)
(10-6 °C)
103
116
477
20.5
Cutting Tool
HSS inserts have been used as cutting tools in 3D orthogonal analyses. The tool has a rake angle
(γ)= -6° and clearance angle (α)= 0° with the edge radius (r) of 0.02 mm because edge radius is
used instead of tool nose radius in 2D orthogonal turning (Fig. 1). The material parameters for
HSS cutting tool was given in Table 3.
Tablo 3. The material parameters for HSS tool
Density
Poisson
(g/cm3) ratio
7.8
0.3
Young's
Thermal
Specific
Thermal
Modulus
Conductivit
Heat
Expansion
(GPa)
y (W/m.K)
(J/kg°C)
(10-6 °C)
207
24
420
11.9
266
Fig. 1 2D model of cutting tool
Finite Element Simulations
The finite element analyses of cutting forces were performed with Advantedge software
depending on finite element method. Advantedge uses an Arbitrary Lagrangian solver and it
has adaptive remeshing function to provide more accurate results although it takes more time.
The first stage of simulation is to determine the workpiece length (3 mm) and height (1 mm)
with workpiece material. The second stage is determining the tool parameters (rake angle,
clearance angle and edge radius) with tool material. The final stage is to enter required
simulation parameters such as feed rate, depth of cut, length of cut and cutting speed after
meshing parameters and coefficient of friction is adjusted.
The interface between tool and work piece was modelled with a standard Coulomb
friction which is assumed as 0.6. The meshing parameters were used as 0.1 mm and 0.02 mm
for maximum and minimum element size, respectively. After these assumptions, a verification
simulation was done with parameters used by Table 1-3. The finite element model was then
verified due to the difference between cutting forces in experiment and simulation is less than
%5. The cutting parameters chosen for these finite element analyses were given in Table 4.
Tablo 4. Cutting Parameters
Levels
Feed
1
2
rate 0.12
Cutting speed 840 1000
Depth of cut 0.10 0.13
Coefficient
of 0.5
friction
267
3
1280
0.16
The workpiece was cut off 2mm in analyses of cutting forces. After the cutting process is
finished, both the chip and tool are removed, and the workpiece is allowed to thermomechanically relax. The simulation model and cutting scheme were shown in Fig. 2.
Fig. 2 2D Simulation Model
Results And Discussion
Only 2D simulation was performed in this study instead of 3D simulation due to low calculation
of time. As shown from Fig 3-4, the influences of cutting parameters (depth of cut and cutting
speed) on cutting forces were assessed by means of the figures. In general, Fig. 3-4 display
different tendency. When both methods were compared, increase in depth of cut generally led
to decrease and then increase in feed force. This situation may be referred to increasing
ploughing effect with increasing depth of cut.
Fig. 3. The Variations of The Feed Force
268
Fig. 4. The Variations of The Main Cutting Force
Numerical Ff values was determined as 4.66% different from that of the experimental
results. This result may arise from the data in literature of the Johnson-Cook model for
workpiece and coefficient of friction in tool-workpiece interface. It was referred in literature
[12,13] that the metallurgical structure and the chemical composition of the standard
manufactured material may be different. Mechanical and physical properties that generate
material models should be determined according to the related experimental workpiece in
cutting simulations. Figure 3 also showed that the cutting speed has great importance on feed
force. The numerical feed force decreased about 34% and 32% while depth of cut increased
from 0.1 to 0.13 mm and from 0.13 to 0.16 mm, respectively by kept feed rate and cutting speed
constant (Figure 3). Decreasing and increasing ratio of numerical feed force was about 38%
and 22% while cutting speed increased from 840 to 1000 m/min and from 1000 to 1280 m/min,
respectively at constant depth of cut and feed rate. The lowest numerical feed force was
obtained as 1.3 N by the feed rate of 0.12 mm/rev, depth of cut of 0.16 mm and cutting speed
of 840 m/min.
It was determined that both experimental and numerical Fc values show similar
tendency (seen Fig. 4) with a percentage of 4.39% difference. The main cutting force values
generally increased with increasing depth of cut while This situation is consistent with literature
[14]. As can be derived from Fig. 4, the numerical main cutting force raised about 3% and 60%
while the depth of cut increased from from 0.1 to 0.13 mm and from 0.13 to 0.16 mm,
respectively at constant feed rate and cutting speed. Increasing ratio of numerical main cutting
force was about 29% and 18% while the cutting speed increased from 840 to 1000 m/min and
from 1000 to 1280 m/min, respectively by constant depth of cut and feed rate. The lowest
269
numerical main cutting force was acquired as 1.3 N with the feed rate of 0.12 mm/rev, depth of
cut of 0.13 mm and cutting speed of 1280 m/min.
Conclusion
In this study, the effects of cutting parameters on cutting forces have been analyzed with finite
element method in hard machining of C23000 brass alloy workpieces with 120 Bhn hardness
using coated carbide tools. 2D orthogonal cutting has been used instead of 3D cutting due to
low calculation of time in finite element analyses. The following conclusions are drawn from
this study:
Firstly, the finite element analysis was resulted in a numerical model and the validation
of model was approved by comparing experimental cutting forces, and power consumption. It
was found that there is an average of 4.66%, and 4.39% deviation between experimental and
simulation results of feed force and main cutting force respectively. It can be concluded that
optimizing the depth of cut, cutting speed and feed rate can result in reduced cutting forces.
Consequently, 2D finite element model can be recommended in machining operations for
optimizing the cutting forces without having the need to perform trial experiments on brass
alloys or their components. Therefore, sustainable machining can be acquired in manufacturing
industry involving easy-to-machine materials through cutting simulations by finite element
method.
REFERENCES
[1] N.M. Vaxevanidis, N.A. Fountas, A. Koutsomichalis, J.D. Kechagias, Experimental
investigation of machinability parameters in turning of CuZn39Pb3 brass alloy, Procedia Structural
Integrity, Volume 10, (2018), Pages 333-341.
[2] M. Hanief, M.F. Wani, M.S. Charoo, Modeling And Prediction of Cutting Forces During
The Turning Of Red Brass (C23000) Using ANN Andregression Analysis, Engineering Science And
Technology, an International Journal 20 (2017) 1220–1226.
[3] Surinder Kumar, Meenu Gupta, P.S. Satsangi, Multiple-Response Optimization of Cutting
Forces İn Turning Of UD-GFRP Composite Using Distance-Based Pareto Genetic Algorithm Approach,
Eng. Sci. Technol. 18 (2015) 1–16.
270
[4] C. Itxaso, J.A. Sarasuaa, Mechanistic Model For Prediction of Cutting Forces İn Turning Of
Non-Axisymmetric Parts, Procedia CIRP 31 (2015) 435–440.
[5] T. Dorlin, F. Guillaume, J.P. Costes, Analysis and Modeling of The Contact Radius Effect
On The Cutting Forces İn Cylindrical And Face Turning Of Ti6al4v Titanium Alloy, Procedia CIRP 31
(2015) 185–190.
[6] J. Xie, M.J. Luo, K.K. Wu, L.F. Yang, D.H. Li, Experimental Study on Cutting Temperature
And Cutting Force in Dry Turning Of Titanium Alloy Using Microgrooved Tool, Int. J. Mach. Tools
Manuf. 73 (2013) 25–36.
[7] D. Philip, P. Selvaraj, P. Chandramohan, M. Mohanraj, Optimization Of Surface Roughness,
Cutting Force And Tool Wear Of Nitrogen Alloyed Duplex Stainless Steel in A Dry Turning Process
Using Taguchi Method, Measurement 49 (2014) 205–215.
[8] S. Sikder, H.A. Kishawy, Analytical Model For Force Prediction When Machining Metal
Matrix Composite, Int. J. Mech. Sci. 59 (2012) 95–103.
[9] John, M.R.S.; Shrivastava, K.; Banerjee, N.; Madhukar, P.D.; Vinayagam, B.K.: Finite
Element Method-Based Machining Simulation for Analyzing Surface Roughness During Turning
Operation with HSS and Carbide Insert Tool, Arab. J. Sci. Eng. 38, 1615–1623 (2013)
[10] M.E. Korkmaz, N. Yaşar, M. Günay, Finite Element Modeling of Residual Stresses and
Cutting Temperature in Hard Turning, International Conference on Engineering and Natural Sciences,
Bosnia and Herzegovina, 2016.
[11] A. Dorogoy and D. Rittel, Determination of the Johnson–Cook Material Parameters Using
the SCS Specimen, Experimental Mechanics 49, 881–885, 2009.
[12] Bil, H.; Kılıç, S.E.; Tekkaya, A.E.: A Comparison of Orthogonal Cutting Data From
Experiments With Three Different Finite Element Models. Int. J. Mach. Tool. Manuf. 44(9), 933-944
(2004)
[13] Özel, T.: The Influence of Friction Models on Finite Element Simulations of Machining.
Int. J. Mach. Tool. Manuf. 46, 518-530 (2006)
[14] Günay, M.; Korkmaz, M.E.; Yaşar, N.: Finite Element Modeling of Tool Stresses on
Ceramic Tools in Hard Turning. Mechanika. 23(3), 432-440 (2017).
271
Evaluation of Critical Factors in Industry 4.0 Transition Processes by R’WOT Analysis
Gizem ACAR1, İlker ÖNALAN2
1
2
Department of Industrial Engineering Karabük University,
Department of Mechatronics Engineering, Karabük University, TURKEY
Abstract: Industry 4.0 was first announced to the public by the German Federal Government
at the 2011 Hannover Fair. The concept of Industry 4.0 is taken into consideration especially in
our country, considering the rapidly developing countries. In this study the work of leading
organizations in Turkey is being examined and the approaches to Turkey’s Industry 4.0 are
reported. In the study, SWOT Analyzes are compiled from the results of civil society
organizations reports, academic publications, public institutions report, consulting companies
and related books. As a result of this review, a single SWOT analysis consisting of 9 items in
each item is presented with the text mining method. The SWOT Analysis obtained is linearly
scored between 1 and 9 points given from the sources using the R'WOT Analysis method.
According to the results of the R'WOT Analysis, our country's approach to Industry 4.0 is
considered as an opportunity of 31%.
Keywords: Industry 4.0, SWOT Analysis, R’WOT Analysis, Ranking Technique, Linear
Combination Technique
REFERENCES
[1] Adıgüzel, Ö. Y. (2016). Endüstri 4.0 ve Nesnelerin İnterneti.
[2] SIEMENS. (2016). Endüstri 4.0 Olma Yolunda: Dijital Fabrikalar.
[3] Aktan, Ç. C. (2003). Değişim Çağında Yönetim. İstanbul Ticaret Odası.
[4] Aktan, Ç. C. (1999, Şubat). 2000’li Yıllarda Yeni Yönetim Teknikleri 2. Türkiye Genç İş
adamları Derneği.
272
[5] Wheelen, T. L., & Hunger, J. D. (1992). Strategic Management and Business, Policy,
Addison-Wesley Publishing Company, Fourth Edition. New York.
[6] Torlak, Ö., Altunışık, R., & Özdemir, Ş. (2002). Modern Pazarlama. İstanbul: Değişim
Yayınları.
[7] Yılmaz, E. (2006). R’wot Tekniği; Arıcılık Sektöründekatılımcı Yaklaşım ile Örnek Bir
Uygulaması. Çevre ve Orman Bakanlığı, Tarsus. 17.
[8] Dönmez, Y., & Gökyer, E.&Aşkın, K.F. (2015) Safranbolu Yörük Köyü ve Yakın
Çevresinin Ekoturizm Potansiyelinin R'WOT Analizi ile Değerlendirilmesi.
[9] Schomoldt, D. L., Peterson, D. L., & Smith, R. L. (1995). The Analytic Hierarchy Process
and Participatory Decision Making. 129-1.
273
Determination of Variables That Affect the Satisfaction Levels of Visiting Tourists By
Logistics Regression Analysis
Fatma ATEŞ1, Nuray TÜRKER2, Filiz ERSÖZ3
1
Sustainable Management and Planning of Natural Resources Department
2
3
Department of Gastronomy and Culinary Arts
Department of Industrial Engineering, Karabük University, TURKEY
Abstract: Tourism in the natural world has an important place. Tourism revenues have a
significant share in the development and development of countries. The economic benefit to be
achieved; it is possible to maximize the natural, historical and cultural beauties. In this study,
which was conducted to determine the satisfaction of tourists, survey was applied to 169
tourists. As a result of analysis; It was concluded that variables such as age, gender, nationality
and income did not affect tourist satisfaction, variables related to purchased products, shops
and sellers had a significant relationship between tourist satisfaction.
Keywords: Tourism Satisfaction Level, Logistic Regression Analysis
Giriş
Turizm, bugün yabancı para girdisini arttıran ve istihdam sağlama nitelikleriyle ülke
ekonomisine katkısı olan, milletlerarası kültürel ve sosyal iletişimi sağlayıcı ve milli ekonomiye
faydalı, milletlerarası iletişimi geliştirici etkileriyle küresel barışın muhafaza edilmesine önemli
katkısı olan bir faaliyettir.
Turizmi; insanların yaşam alanlarının dışına çıkıp, farklı mekanlara gidip farklı faaliyetlere
katılmak isteğiyle, burada konaklayarak gerçekleştirdikleri geçici hareketlilik olarak
tanımlayabiliriz. Gerçekleştirilen bu faaliyetler; konaklama süresine, gerçekleştirme amacı ve
şekline, ulaşım çeşidine, maliyetine, yolcu sayısına göre çeşitli kategorilere ayrılmıştır 20.
Yüzyılın başlarından sonraki dönemlerde hızlı bir ilerleme kaydetmiş ve bütün dünyaya
274
yayılmıştır. Turizm küresel ekonomiye yön veren bir sektör olarak karşımıza çıkmakta ve her
geçen gün önemi artmaktadır.
Dünya genelinde, sürekli değişen tüketici ihtiyaçlarına bağlı olarak farklılık gösteren
turizm hareketleri ve turistlerin görüşleri, destinasyonların gelişmişlik seviyelerini de
etkilemektedir. Turizm faaliyetlerine katılan bireyler, tatillerini geçirecekleri destinasyonu
seçerken, o yerin doğal kaynaklarına, somut ya da somut olmayan kültürel miras varlıklarına,
altyapı ve üstyapı olanaklarına, bölge veya yörede doğrudan ya da dolaylı olarak turizm sektörü
içerisinde yer alan paydaşların birbirleriyle olan ilişkilerine ve sunulan hizmetlerin kalitesi gibi
çeşitli unsurlara değişik seviyelerde önem vermektedirler. Potansiyel turistlerin, tatil öncesi
önem verdikleri bu unsurlardan, tatilleri sonrası memnun kalma seviyeleri, gelecek yıllarda
tekrar söz konusu destinasyona gitmelerini ya da destinasyonun gönüllü birer pazarlamacısı
haline gelmelerinde en kilit noktalardan birisidir [14].
Sürdürülebilirlik prensiplerine bağlı politikalar hayata geçirilemediği takdirde kültürel
turizm, bölgelerin doğal ve beşeri turizm varlıkları üzerinde telafisi olmayan etkiler
bırakacaktır. Kültür turizmini geliştirmek ve gelecek nesillere de miras bırakmak için atılacak
adımlar, mahalli ve folklorik değerlerin yok olmasını engelleyecektir.
Ülkemiz coğrafi konum olarak eski dünya karalarının bitişme noktasında bulunur. Bu
durumda karalar arasındaki etkileşimlerin ülkemiz topraklarında gerçekleşmesi olağandır. Aynı
zamanda Türkiye, medeniyetler arasında cereyan eden etkileşiminle birçok medeniyete ev
sahipliği yapmıştır. Çeşitli savaşlara, göçlere ev sahipliği etmiş halen de bulunduğu coğrafyanın
handikaplarını kültürel ve ekonomik olarak hissetmektedir. Doğal ve kültürel zenginlikleriyle
ciddi anlamda turizm potansiyeline sahip bir ülkedir. Sahip olduğumuz zenginliklerin
sürdürülebilir kulllanımını sağlamak bu potansiyeli etkin ve verimli kullanmanın yegâne
koşuludur.
Türk ekonomisinin de vazgeçilmez temel taşlarından birisi olan turizm, bugünkü dış ticaret
açığına, enflasyona ve işsizliğe çare arayan hükümetlerin önemle üzerinde durduğu bir konudur
[9]. Kılıç ve Pelit (2004); Seddighi ve Theocharous (2002) atfen turizm faaliyetlerinin bir
bölgede ya da destinasyonda gelişebilmesinde çeşitli unsurlar etkili olmadığını, bölgenin doğal
yapısı, sosyal yapısı, alt ve üst yapısı, ulaşımı, bölgedeki yerel yönetimlerin tavrı, halkın
275
turizme ve turistlere bakış açısı, bölgede faaliyet gösteren işletmelerin tutumları bölge
turizminin gelişmişliğini etkileyen önemli unsurlar olduğunu ifade etmiştir [8].
UNESCO Dünya Miras Listesi’nde ‘’En iyi korunmuş 20 şehir’’ arasında bulunan
Safranbolu, gerek Osmanlı-Türk mimarisiyle inşa edilmiş meskenleri gerekse; hamam, han ve
çeşmelerin meydana getirdiği kültürel atmosferiyle yerli yabancı çok sayıda turiste ev sahipliği
yapmaktadır. Kültür turizmi açısından ülkemizin en fazla ziyaret edilen şehirlerinden bir
tanesidir. Burada yaşayanlar insanlarında önemli bir kısmı geçimini turizmden elde etmektedir.
Kentte birçok hediyelik eşya dükkânı, yeme içme yerleri ve eğlence merkezleri bulunur.
Dönemsel olarak maddi getirisinde değişikler olmakla birlikte özellikle yabancı turistlerin sık
ziyaret ettiği aylarda esnaflar tarafından önemli miktarda kazanç elde edilmektedir. Belediye
tarafından düzenlenen festivaller ve etkinliklerinde turizm gelirleri üzerindeki payı büyüktür.
Literatür Araştırması
Müşteri Memnuniyetini Oluşturan Faktörlerin Müşteri Sadakatine Etkisinin Lojistik Regresyon
Analizi ile İncelenmesi adlı [18] çalışmada, İzmir ve Afyonkarahisar’daki beş yıldızlı termal
otel işletmelerinde konaklayan müşterilerin memnuniyetlerini oluşturan faktörlerin sadakatleri
üzerindeki etkisinin lojistik regresyon analizi ile belirlenmesidir. Bu kapsamda İzmir ve
Afyonkarahisar’daki beş yıldızlı termal otel işletmelerinde konaklayan 423 müşteriye anket
uygulaması yapılarak veriler elde edilmiştir. Verilerin analizi sonucu, müşteri memnuniyeti ile
müşteri sadakati arasındaki pozitif yönlü kuvvetli bir ilişki olduğu tespit edilmiştir. Uygulanan
lojistik regresyon analizi sonuçlarına göre ise, odalar bölümü hizmetlerinden memnun olan
müşterilerin memnun olmayan müşterilere göre sadakat davranışlarının 3,5 kat, yiyecek içecek
bölümü hizmetlerinden memnun olan müşterilerin memnun olmayan müşterilere göre 6,7 kat
ve genel değerlendirme sonucu memnun olan müşterilerin memnun olmayanlara göre 3,4 kat
daha fazla olduğu tespit edilmiştir [18].
E-hizmet kalitesinin incelendiği çalışmada İnternet perakendeciliğinde algılanan hizmet
kalitesinin (e-hizmet kalitesi) müşteri memnuniyeti (e-memnuniyet) üzerindeki etkisini tespit
etmek ve bu etkinin farklı sektörlere göre nasıl değiştiğini ortaya koymak amaçlanmaktadır. Bu
kapsamda iki farklı sektör (hazır giyim ve kitap) seçilmiş ve sektördeki en büyük iki rakip
markanın müşterileri araştırma kapsamına alınmıştır. İnternet üzerinden anket yöntemi ile 590
276
kişiden elde edilen verilere regresyon analizi uygulanmıştır. Araştırmada e-hizmet kalitesi
algısının e-memnuniyet üzerinde anlamlı bir etkiye sahip olduğu bulunmuştur. e-hizmet kalite
algısında önemli unsurlardan olan "gizlilik" ve "teknik" boyutlarının, e-memnuniyeti
açıklamakta anlamlı bir etkisinin olmadığı, "etkinlik", "işlem gerçekleştirme", "müşteri
hizmetleri", "tasarım", ’’eğlence” boyutlarının etkilerinin ise anlamlı olduğu bulunmuştur. En
önemli etkinin ise "işlem gerçekleştirme" boyutunda olduğu, en az etkinin ise "tasarım"
boyutunda olduğu saptanmıştır. Ayrıca e-hizmet kalitesinin e-memnuniyet üzerindeki
etkilerinde sektörlere göre kısmi farklılıkların olduğu tespit edilmiştir [6].
Müşteri memnuniyetine ilişkin çalışmada, algılanan fiyat, algılanan değer ve algılanan
faydanın tekrar satın alma eğilimi üzerinde doğrudan ve dolaylı etkileri araştırılmıştır.
Çalışmanın verileri şehirlerarası yolcu taşıyan bir firmanın 182 müşterisinden elde edilmiştir.
Değişkenler arasındaki ilişkiyi bulabilmek için yapısal eşitlik modeli kullanılmıştır [20].
Müşteri memnuniyetine ilişkin bir diğer çalışmada çeşitli demografik faktörlerden nasıl
etkilendiğini analiz etmek amacıyla yapılmıştır. Araştırma, Ankara ilinde bulunan bir kamu
hastanesinden sağlık hizmeti alan hastalar üzerinde yapılmış olup tanımlayıcı niteliktedir.
Araştırma kapsamındaki hastaların müşteri memnuniyetini (hasta memnuniyeti) ölçmek için
T.C. Sağlık Bakanlığı’nın resmi internet sitesinde yer alan memnuniyet anketleri yapılmıştır.
148 adet anket formu değerlendirilmiştir [7].
Tramvay Yolcu Memnuniyetinin Lojistik Regresyon Analiziyle Ölçülmesi: Estram
Örneği adlı çalışmasında; toplu taşıma araçlarından biri olan tramvaya yönelik yolcu
memnuniyeti, Eskişehir tramvay sistemi (Estram) örneğinde, Binominal Lojistik Regresyon
Analizi ile incelenmektedir. İki üniversiteye sahip olan Eskişehir’de öğrenci nüfusun fazla
olması ve tramvay için önemli bir yolcu kitlesi olacakları düşüncesiyle çalışma, her iki
üniversiteden
basit
tesadüfî
örnekleme
yoluyla
seçilen
300
öğrenci
üzerinde
gerçekleştirilmiştir. Öğrencilerin memnuniyetleri ile ilgili binominal düzeyde, gizil bir
değişken kullanılmıştır. Uygulanan Binominal Lojistik Regresyon Analizi sonucunda;
öğrencilerin Estram’dan memnuniyetleri üzerinde modele alınan tüm bağımsız değişkenlerin
negatif etkileri olduğu belirlenmiştir [12].
277
Müşteri Memnuniyetinin tahmini için yapılan çalışmada Yapay Sinir Ağları, Lojistik
Regresyon ve Ayırma Analizinin performansları karşılaştırılmıştır. Veriler Uşak’taki kamu
hastanelerinde 2007 yılında yapılan hasta memnuniyetini ölçmeyi amaçlayan bir anket
uygulanarak elde edilmiştir ve 364 hastayı kapsamaktadır. Sonuçlar Yapay Sinir Ağlarının
diğer yöntemlere göre müşteri memnuniyetini daha iyi tahmin ettiğini göstermiştir [21].
Turist memnuniyetine ilişkin yapılan çalışmada, Alanya ilçesinin Türk Turizmi
içerisindeki önemini belirtmek, dış turizmin bölgede meydana getirdiği ekonomik etkileri
açıklamak ve konaklama işletmelerinin turist memnuniyeti üzerindeki etkisi incelenmiştir.
Araştırmada yerli ve yabancı turistlere 27 sorudan oluşan bir anket uygulanmıştır. Toplam 104
anket değerlendirmeye alınmış ve SPSS 17 programında güvenilirlikleri kontrol edilip, faktör
analizi, korelasyon analizi, regresyon analizi yöntemleri ile incelenmiştir [11].
Marina işletmelerinde yapılan bir diğer çalışmada, ilişkisel pazarlama uygulamalarının
tekrar satın alma niyeti, tavsiye etme niyeti ve yönetimden memnuniyet düzeyi üzerinde bir
etkisinin bulunup bulunmadığı belirlemek ve marina yöneticilerine bu konuda yol gösterecek
katkı ve önerilerde bulunulmuştur. Bu amacı gerçekleştirmek üzere Antalya Bölgesi’ nde
faaliyet gösteren bir marina işletmesinin 78 müşterisine yüz yüze anket uygulaması
gerçekleştirilmiş ve veriler SPSS paket programıyla analiz edilmiştir [3].
Materyal ve Yöntem
Safranbolu’yu ziyaret eden turistlerin memnuniyet düzeylerini belirlemek amacıyla Çinli ve
Tayvanlı olmak üzere 169 yabancı turiste anket uygulanmıştır. Bu amaca yönelik
memnuniyetin hangi değişkenlere bağlı olarak ortaya çıktığını görebilmek için lojistik
regresyon analizinden yararlanılmıştır.
Lojistik regresyon; cevap değişkenin (y) kategorik olarak, ikili (binary, dishotomous) ve çoklu
(multinominal) kategorilerde gözlendiği durumlarda açıklayıcı değişkenlerle (xi, =1,2,…,k)
sebep sonuç ilişkisini belirlemede yararlanılan bir yöntemdir. Cevap değişkenin (y) değişimi
üzerinde etkili olan açıklayıcı değişkenlerin/risk faktörlerinin (x,) etki büyüklüklerini (Odds
Radio) belirlemeyi sağlayan bir yöntemdir. Ayrıca Lojistik regresyon, açıklayıcı
değişkenlere/risk faktörlerine göre cevap değişkenin beklenen değerlerinin olasılık olarak elde
edildiği sınıflama ve atama işlemi yapmaya yardımcı olan bir yöntemdir [17].
278
Lojistik Regresyon (LR) Analizi; sürekli ya da kesikli ya da her ikisinin bir arada
bulunduğu sürekli ve kesikli değişkenler setinden yola çıkarak kesikli bir sonucun
yordanmasını sağlar. Sosyal bilimlerde sıklıkla kullanılan çok yönlü frekans analizi ve çoklu
regresyon analizlerinden farklı olarak LR analizi, bağımlı değişkenin kesikli olabilmesine
imkân tanımaktadır. Bu bakımdan sonuçların kategorik olduğu durumlar için bu analizin yaygın
kullanımı söz konusudur.
Çok yönlü frekans analizinde yordayıcıların kesikli olması gerekirken LR analizinde
böyle bir zorunluluk bulunmamakta, yordayıcılar sürekli olabildiği gibi kesikli de olabilmekte
ya da hem sürekli hem kesikli yordayıcılar bir arada bulunabilmektedir. Çoklu regresyon analizi
ile karşılaştırıldığında ise LR analizi negatif kestirim olasılıkları üretmemektedir. Ayrıca LR
analizi, çoklu regresyon analizi ve diskriminant analizinden farklı olarak yordayıcı
değişkenlerin dağılımlarına yönelik varsayımlara sahip değildir. Bu bakımdan LR analizi,
bahsedilen diğer yöntemlere göre çok daha esnektir [23].
Verilerin
analiz
edilmesinde
multinominal
lojistik
regresyon
analizinden
yararlanılmıştır. Çalışmada kullanılan verilerin analiz edilmesinde IBM SPSS 17.0 paket
programı kullanılmıştır.
Lojistik Regresyon Analizi, bağımlı değişkeni iki veya ikiden çok kategoriye sahip olan
bir denklemde, bağımsız değişken veya değişkenler ile bağımlı değişken arasındaki ilişkiyi
ifade etmekte kullanılan bir yöntemdir. Ancak burada Regresyon Analizinden farklı olarak
bağımlı değişkenin kategorilere sahip olması sebebiyle bağımsız değişkenin bağımlı değişken
üzerindeki etkisi olasılık olarak ifade edilir [19]. Lojistik regresyon analizinin temel odağı,
bireylerin hangi grubun üyesi olduğunu kestirmede bir regresyon denklemi oluşturmaktır. Bu
çalışmada, iki kategorili (ikilem/dichotomous/binary) bağımlı değişken olarak ifade edilen
belirli gruplara üye olma durumunu en iyi açıklayan bağımsız değişkenler kombinasyonunu
belirlemeye yönelik ikili lojistik regresyon analizinin (binary logistic regression analysis) temel
kavram ve süreçlerini açıklamak amaçlanmaktadır [10].
Herhangi bir regresyon modelinde E(Y/x) ifadesi koşullu ortalama değerini
göstermektedir. Bu ifade, bağımsız değişken (x) verildiğinde bağımlı değişkenin (Y) ortalama
değerini göstermektedir [13].
279
E(Y/x)= β0 + β1X
Lojistik regresyonda ise x verildiğinde Y’nin koşullu dağılım ortalaması 𝜋(𝑥) = 𝐸(𝑌/𝑥)
olarak gösterilir. Lojistik regresyon modelinin özel durumu;
E(Y/x) = π(x) =
şeklindedir. Yukarıdaki eşitliğe lojit dönüşüm uygulandığında aşağıdaki eşitlik elde edilir [13].
g(x) = ln
( )
( )
= β0 + β1X
Logit değeri ile bağımsız ve bağımlı değişkenler arasında doğrusal bir ilişki vardır. Bu
değer -∞ ile +∞ arasında olabilir [4].
İkili lojistik regresyon modelinde olayın gerçekleşme ve gerçekleşmeme durumu 0 ve
1 olmak üzere iki durum söz konusudur. Olayın gerçekleşme olasılığının, gerçekleşeme
olasılığına oranı odds oranı olarak tanımlanmaktadır. Odds oranı 0 ile +∞ arasında değer
alabilmektedir [1].
Odds Oranı =
( )
( )
=e β0 + β1X
Lojistik regresyon analizinde regresyon katsayılarının tahmininde en küçük kareler
yöntemi yerine, en çok olabilirlik yöntemi kullanılmaktadır. En çok olabilirlik yönteminde bir
olayın olma olasılığının maksimum olması istenir [2].
Verilere basit ya da çoklu regresyon analizlerinin uygulanabilmesi için değişkenlerin
bazı koşullara (varsayımlara) uyması gerekir. Bu koşulların sağlanamadığı veri setlerine basit
ya da çoklu regresyon analizleri uygulanamaz. Lojistik regresyon analizi ise Normal Dağılım
Varsayımı, süreklilik varsayımı ön koşulu gerektirmeyen bir regresyon yöntemidir [17].
Uygulama
Bu bölümde, kullanılan veriler, çalışmanın amacı ve lojistik regresyon analizi sonucunda elde
edilen bulgular yorumlanmaya çalışılmıştır. Bireyin eğitim durumunun, istihdam durumunun,
yaşının, sağlık ve hastalık durumunun maddi yoksunluğu anlamlı bir şekilde etkileyip
etkilemediği belirlenmek istenmiştir. Bunların yanı sıra konutun oda sayısının, ısıtma
280
sisteminin maddi yoksunluğu belirlemede anlamlı bir etkisi olup olmadığı tespit edilmeye
çalışılmıştır. Bu varsayımlar ışığında Lojistik regresyon analizi kurularak sonuçlar
değerlendirilmiştir.
Araştırmanın temel hipotezi: ‘’Bireyin milliyeti, cinsiyeti, geliri, yaşı, ürünün kalitesi, ürünün
çeşitliliği, satıcının sunduğu hizmet ve indirimli fiyat değişkenleri memnuniyet düzeyini
anlamlı bir şekilde etkileyeceği’’dir.
Alt hipotezler ise;
H1: Milliyet değişkeni memnuniyet düzeyini belirlemede anlamlı bir etkiye sahiptir.
H2: Cinsiyet değişkeni memnuniyet düzeyini belirlemede anlamlı bir etkiye sahiptir.
H3: Gelir değişkeni memnuniyet düzeyini belirlemede anlamlı bir etkiye sahiptir.
H4: Yaş değişkeni memnuniyet düzeyini belirlemede anlamlı bir etkiye sahiptir.
H5: Ürünün kalitesi değişkeni memnuniyet düzeyini belirlemede anlamlı bir etkiye sahiptir.
H6: Ürünün çeşitliliği değişkeni memnuniyet düzeyini belirlemede anlamlı bir etkiye sahiptir.
H7: Satıcının sunduğu hizmet değişkeni memnuniyet düzeyini belirlemede anlamlı bir etkiye
sahiptir.
H8: İndirimli fiyat değişkeni memnuniyet düzeyini belirlemede anlamlı bir etkiye sahiptir.
Çalışmada ele alınan değişkenler ve alt kategoriler Tablo 1’de sunulmuştur.
Tablo 1: Araştırmada Kullanılan Değişkenler
Bağımsız Değişkenler
1. Milliyet
2. Cinsiyet
3. Gelir
Alt kategoriler
1. Çin
2. Tayvan
1. Kadın
2. Erkek
1. 1000 TL’den az
2. 1001-1500 TL
3. 1501-2000 TL
4. 2001-2500 TL
281
4. Yaş
5. Ürün kalitesi
6. Ürün çeşitliliği
7. Satıcının sunduğu hizmet
8. İndirimli fiyat
5. 2501-3000 TL
6. 3001-3500 TL
7. 3501-4000 TL
8. 4001-4500 TL
9. 4501-5000 TL
10. 5000 TL ve üzeri
1. 20 yaş ve altı
2. 21-30 yaş
3. 31-40 yaş
4. 41-50 yaş
5. 51-60 yaş
6. 61 yaş ve üstü
1. Hiç memnun değilim
2. Memnun değilim
3. Kısmen memnunum
4. Memnunum
5. Çok memnunum
1. Hiç memnun değilim
2. Memnun değilim
3. Kısmen memnunum
4. Memnunum
5. Çok memnunum
1. Hiç memnun değilim
2. Memnun değilim
3. Kısmen memnunum
4. Memnunum
5. Çok memnunum
1. Hiç memnun değilim
2. Memnun değilim
3. Kısmen memnunum
4. Memnunum
5. Çok memnunum
Ele alınan bağımsız değişkenler yardımıyla memnuniyet düzeyi bağımlı değişkenini
belirlemek amacıyla Lojistik Regresyon Modeli kurulmuştur. Öncelikle modelin genel anlamda
uygunluğunu gösteren Omnibus test sonuçları sunulmuştur.
Tablo 2: Model Katsayılarının Genel Testi
Ki-Kare
sd
Sig
Adım 1 Adım
158,636
19
,000
Blok
158,636
19
,000
Model
158,636
19
,000
282
Bütün değişkenler modele eklendikten sonra model uyum iyiliği için Ki-Kare değerine
bakılır. Modelin genel anlamlılığının, yani uyum iyiliğinin istatistiksel olarak anlamlı olduğu
görülmektedir (p<0,01). Modelin verilere uygunluğunu ve modelin genel uyumunu gösteren
R2 değeri Tablo 4’te verilmiştir.
Tablo 3: Hosmer ve Lemeshow Testi
Adım
Ki-Kare
df
Sig
1
13,680
8
,091
Tablo 4: Model Özeti
Adım
-2Log likelihood
Cox & Snell R Square
Nagelkerke R Square
1
39,237
,609
,883
Hosmer ve Lemeshow test sonucuna göre; tahmin edilen lojistik regresyon modelinin
verilere uygun olduğu (p=0.091) görülmüştür. Cox & Snell R2 ve Nagelkerke R2 değerleri,
model tarafından bağımlı değişkende açıklanan varyansın büyüklüğünü göstermektedir.
Modelin genel uyumunun iyi olduğu (Cox & Snell R2=0,609; Nagelkerke R2=0,883)
belirlenmiştir. Maddi yoksunluğun toplam değişimin %88’i ele alınan bağımsız değişkenler
tarafından açıklanmaktadır. İkili Lojistik regresyon analizine ait katsayı tahminleri ve odds
oranları Tablo 5’da sunulmuştur.
Tablo 5: Model Tahmin Sonuçları
Bağımsız Değişken
Kategoriler
Score
df
Sig
Milliyet
,000
1
,989
Cinsiyet
,000
1
,989
1000 TL’den az
4,983
9
,836
1001-1500 TL
,148
1
,701
1501-2000 TL
1,136
1
,287
2001-2500 TL
,487
1
,485
2501-3000 TL
,996
1
,318
3001-3500 TL
,117
1
,732
3501-4000 TL
,376
1
,540
Gelir
283
4001-4500 TL
1,142
1
,285
4501-5000 TL
1,142
1
,285
,021
1
,885
20 yaş ve altı
4,850
5
,434
21-30 yaş
3,011
1
,083
31-40 yaş
,323
1
,570
41-50 yaş
1,493
1
,222
51-60 yaş
,016
1
,898
61 yaş ve üstü
,577
1
,448
Ürün Kalitesi
62,334
1
,000
Ürün Çeşitliliği
80,577
1
,000
Satıcının Sunduğu Hizmet
65,587
1
,000
İndirimli Fiyat
46,069
1
,000
5000 TL ve üstü
Yaş
p<0,05
Turist memnuniyetini etkileyen değişkenlerin analizi yapıldığında Tablo 5’e göre
memnuniyeti etkileyen değişkenlerden ürün kalitesi, ürün çeşitliliği, satıcının sunduğu hizmet
ve indirimli fiyat ile memnuniyet arasında anlamlı bir ilişki bulunmaktadır. Bu sonuca göre alt
hipotezlerden H5, H6, H7, H8 ‘in gerçekleşmiş olduğu söylenebilmektedir.
Tablo 6: Sınıflandırma Tablosu
Tahmin Edilen
Memnun Değil
Memnun
Doğruluk (%)
Memnun Değil
0
46
,0
Memnun
0
123
100,0
Genel %
72,8
Kurulan lojistik regresyon modelinin sınıflandırma tablosu Tablo 6’da verilmiştir. Testin
duyarlılık oranı %100 olarak elde edilmiştir. Bu durum gerçek durumda turistlerin memnun
olanlarının %100’ünün doğru olarak tahmin edildiğini göstermektedir. Modelin doğru
sınıflandırma oranı anket uygulanan turistlerin %72,8’inin memnuniyet durumunu doğru
tahmin ettiğini göstermektedir. Modelin sınıflandırma gücünün oldukça iyi olduğu söylenebilir.
284
Sonuç ve Öneriler
Turistlerin milliyetinin, cinsiyetinin, yaşının, maddi gelirinin, ürün kalitesinin, ürün
çeşitliliğinin, satıcının sunduğu hizmetin ve indirimli fiyatın turist memnuniyetlerini anlamlı
bir şekilde etkileyip etkilemediği belirlenmek istenmiştir. Bu varsayımlar ışığında lojistik
regresyon analizi kurularak sonuçlar değerlendirilmiştir.
Bu çalışmada Safranbolu’yu ziyaret eden turistlerin memnuniyet düzeylerini etkileyen
değişkenlerin neler olduğunu ortaya koymak amaçlanmıştır. Yapılan analizler neticesinde turist
memnuniyetini etkileyen faktörlerden ürün kalitesi, ürün çeşitliliği, satıcının sunduğu hizmet
ve indirimli fiyat değişkenlerinin anlamlı olduğu anlaşılmıştır.
Yerel
ekonominin atılımı, herhangi
bir
bölgeye
has
değişim
süreçlerinin
motivasyonudur şeklinde açıklanabilir. Turizm, ekonominin önemli sektörlerinden biridir.
Geçmişten günümüze kadarki süreçte Safranbolu kültürel özellikleri, tarihi ve otantik kendine
has mimarisiyle her daim yerli ve yabancı turistlerin ziyaret ettiği önemli merkezlerden biri
olmuştur. Şehir, 1994 yılında UNESCO Miras Listesine dahil edilerek koruma altına alınmıştır.
Uluslararası koruma statüsü kazanarak tanınırlığı da artan Safranbolu’yu görmek için gelen
turist sayısında da artış görülmüştür. Safranbolu’nun şehir kimliğinin vurgulanarak turizme
katkı sağlanması, iç turizmde hatrı sayılır derecede isim yapmasına vesile olmuştur.
Sürdürülebilir turizm uygulamaları, yerel ekonomi canlandırılabilmekte ve yabancı
turist çekiciliğinin yanı sıra iç turizmi de hareketlendirebilmektedir. Bu tür çalışmalarla turist
memnuniyetinin hangi faktörlere göre değişiklik gösterdiği belirlenerek tarihi ve kültürel
zenginliklerimiz için planlanan politikaların o yönde geliştirmek hem elde edilen turizm
gelirleri için hem de yörede yaşayan insanların ve çevrenin kalkındırılması açısından isabetli
olacaktır.
KAYNAKÇA
[1] Agresti, A. (1996). An Introduction To Categorical Data Analysis (Vol. 1350), New York:
Wiley.
[2] Alpar, R. (2013). Çok Değişkenli İstatistiksel Yöntemler (4. Baskı), Detay Yayıncılık,
Ankara.
285
[3] Arlı, E. (2013) Marina İşletmeciliğinde İlişkisel Pazarlama Uygulamalarının Tekrar Satın
Alma Niyeti, Tavsiye Etme Niyeti ve Memnuniyet Üzerindeki Etkisi. Anadolu Üniversitesi Sosyal
Bilimler Dergisi.
[4] Bayram, N. (2004). Multinominal Lojistik Regresyon Analizinin İstihdamdaki İşgücüne
Uygulanması, İktisat Fakültesi Mecmuası, 54(2): 61- 75.
[5] Bilgin, Y. (2017) Restoran İşletmelerinde Hizmet Kalitesi, Müşteri Memnuniyeti ve Müşteri
Sadakatinin Ağızdan Ağıza Pazarlamaya Etkisi. İşletme Araştırmaları Dergisi, Bartın Üniversitesi
İktisadi ve İdari Bilimler Fakültesi, Turizm İşletmeciliği Bölümü, Bartın.
[6] Bozbay, Z., Yaman, Y., & Özkan, E. (2016). The Role of Service Quality on Customer
Satisfaction in Internet Retailing: A Comparative Study of Apparel and Book Industries. Journal of
Transportation and Logistics, 1(1), 19–19.
[7] Cihangiroğlu, N. Uzuntarla, Y. (2016) Müşteri Memnuniyetinin Çeşitli Demografik
Özellikler Açısından Analizi: Bir Kamu Hastanesi Örneği.
[8] Çakıcı, A, C. Aksu, M. (2006) Gökçeada’ya Gelen Turistlerin Beklenti ve Tatmin
Düzeylerinin Karşılaştırılması. İşletme Fakültesi Dergisi.
[9] Çımat, A., & Bahar, O. (2003). Turizm Sektörünün Türkiye Ekonomisi İçindeki Yeri ve
Önemi Üzerine Bir Değerlendirme. Akdeniz İİBF Dergisi, (6), 1–18.
[10] Çokluk. Ö, ‘’2010’’, Lojistik Regresyon Analizi: Kavram ve Uygulama. Ankara
Üniversitesi Eğitim Bilimleri Fakültesi, Ölçme ve Değerlendirme Bölümü, Eğitim İstatistiği ve
Araştırma Anabilim Dalı, Ankara.
[11] Demir, V. Altındağ, E. (2017) Konaklama İşletmelerinde Turist Memnuniyetinin
Değerlendirilmesi: Alanya İlçesi Örneği. Turizm ve Araştırma Dergisi
[12] Girginer, N., & Cankuş, B. (2008). Tramvay Yolcu Memnuniyetinin Lojistik Regresyon
Analiziyle Ölçülmesi: Estram Örne ğ i. Yönetim ve Ekonomi, 15(1), 181–193.
[13] Hosmer, D. W. ve Lemeshow, S. (2000). Applied Logistic Regression., 2nd edn.(Wiley:
New York.). NY, USA.
[14] İpar, M. S. Doğan, M. (2013) Destinasyonun Turist Açısından Önem-Memnuniyet Modeli
ile Değerlendirilmesi: Edremit Üzerine Bir Uygulama. Adıyaman Üniversitesi Sosyal Bilimleri
Enstitüsü Dergisi, Adıyaman.
286
[15] Karcı, Z. Bayram Arlı, N. (2018), Maddi Yoksunluğu Etkileyen Değişkenlerin Regresyon
Analizi ile Belirlenmesi. Süleyman Demirel Üniversitesi, İktisadi ve İdari Bilimler Fakültesi Dergisi,
Isparta.
[16] Meydan Uygur, S. Baykan, E. (2007), Kültür Turizmi ve Turizmin Kültürel Varlıklar
Üzerindeki Etkileri. Gazi Üniversitesi, Ticaret ve Turizm Eğitim Fakültesi, Ticaret ve Turizm Eğitim
Fakültesi Dergisi, Ankara.
[17] Özdamar, K. (2009), Paket Programlar ile İstatistiksel Veri Analizi-1. Kaan Kitabevi,
7.Baskı, s.571. Eskişehir.
[18] Soybalı, H. H., & Cengiz Mutlubaş, I. (2017). Müşteri Memnuniyetini Oluşturan
Faktörlerin Müşteri Sadakatine Etkisinin Lojistik Regresyon Analizi ile İncelenmesi (The Investigation
of Effect of Customer Satisfaction Factors on Customer Loyalty with Logistics Regression Analysis).
Turk Turizm Arastirmalari Dergisi, 1(3), 1–15.
[19] Şentürk. E, ‘’2011’, Mutluluk Düzeyinin Sosyo-Demografik Özelliklerle Lojistik
Regresyon Analizi Aracılığıyla İncelenmesi ve Türkiye İçin Bir Uygulama. Marmara Üniversitesi,
Sosyal Bilimler Enstitüsü, Ekonometri Anabilim Dalı, İstatistik Bilim Dalı, İstanbul.
[20] Tayyar, N. Bektaş, Ç. (2009) Algılanan Değer ve Müşteri Memnuniyetinin Tekrar
Satınalma Eğilimi Üzerine Etkisi.
[21] Tayyar, N. (2010) Müşteri Memnuniyeti Tahmininde Yapay Sinir Ağları, Lojistik
Regresyon ve Ayırma Analizinin Performanslarının Karşılaştırılması. Süleyman Demirel Üniversitesi,
İktisadi ve İdari Bilimler Fakültesi Dergisi, Isparta.
[22] Yıldız, B. Çiğdem, Ş. (2018) Havayolu Hizmet Kalitesinin Müşteri Memnuniyeti
Üzerindeki Etkisinin Yapısal Eşitlik Modeli ile Analizi. Bingöl Üniversitesi Sosyal Bilimler Enstitüsü
Dergisi, Bingöl.
[23] http://www.akademikistatistik.com (Erişim: 15.12.2018).
287
Demand Estimation of Wood Quantity Used in Wood Industries
Sercan ODABAŞ1, Taner ERSÖZ
1Department of Industrial Engineering, 2Department of Actuary and Risk Management,
Karabük University, TURKEY
Abstract: With the rapid development of technology, competition in the markets is increasing.
In the conditions of rising competition, it is inevitable for companies to make predictions for
the future. The accuracy of the forecasts will also facilitate the decision-making of companies
in the future. Demand forecasts are very important for every company considering these
conditions. Demand forecasts play an active role in tactical and strategic decisions taken on the
administrative side of companies to achieve their near or far-term goals, bringing the company's
assets closer to optimal and thus ensuring the link between the targets set and the operations
implemented. According to the data obtained from the firm of Antik Atölye operating in Düzce,
the regression analysis method and weighted moving average method were compared. For both
methods, estimation modeling was established and implemented. As a result for the Antik
Atölye company operating in the wood industry regression analysis method was more effective
than weighted moving average method.
Keywords: Forecasting, Regression Analysis, Ağırlıklı Hareketli Ortalama Yöntemi
Giriş
Ahşap ürünler, dünyada yapı malzemeleri başta olmak üzere kapılarda, yer döşemelerinde,
pencerelerde, çatılarda, masa ve koltuk takımlarında yoğun bir şekilde kullanılmaktadır. Aynı
zamanda kullanım yerlerine göre iç – dış mekan mobilyaları, hareketli veya sabit, sökülebilir,
modüler, fonksiyonel, lamine, modern, klasik mobilyalar olarak gruplara ayrılır. İnsanoğlunun
varoluşundan bu yana sürekli hayatın içerisinde olan ahşap oldukça yüksek katma değere sahip
olup önemli ölçülerde istihdam sağlamaktadır. Ülkemizde ahşap sektörü üç milyar doların
üzerinde katma değer oluşturan bir potansiyele sahiptir. Gelişen teknolojinin sektöre etkisi
kaçınılmaz olup söz konusu sektörde birçok teknolojik cihazlar tezgahlar ve yazılımlar
288
kullanılmaktadır. Bütün veya parçalar halinde yapılan üretim sonucunda görselliği ve
dayanıklılığı artırmak için metal (demir ayak ve tutucular vb.) ve kimyasal (epoksi vb.)
malzemeler kullanılmaktadır.
Hızla gelişmekte olan ahşap sektörü ülkemizde diğer Avrupa ülkeleri ile
kıyaslandığında önemli bir yere sahiptir. Teknoloji yoğun olmasından ziyade emek yoğun olan
bir sektör olması ülkemizi Avrupa ülkeleri karşısında avantajlı konuma getirmektedir.
Hammadde ve işçiliğin Avrupa’ya kıyasla daha ucuz olması imalat yapılabilmesi kısıtları
düşünüldüğünde ihracat dengesine büyük katkı sağlamaktadır.
21. yüzyılın enerji ve çevre yüzyılı olduğu değerlendirildiğinde dayanıklılığı ve
sağladığı istihdam ile ahşap sektörü gelişimini devam ettirmektedir. Ahşap üretimin hızla
büyümesi ekonomik büyümeyi de olumlu etkiletmektedir. Avrupa’ya olan coğrafi yakınlık ve
kolaylıkla ulaşım sağlanan kaliteli hammadde kriterleri doğrultusunda kaliteli işçilik ile
mükemmele yakın üretim yapılması her geçen gün ihracat rakamlarımızı geliştirmekte ve
ekonomik büyümeye katkı sağlamaktadır.
Ahşap sektörü ihracat rakamlarını 90’lı yıllar baz alındığında 2000 yılı başlarında %16,
2010 yılı başlarında %52, 2017 yılı itibari ile de %75 lik artış göstererek ekonomik büyümeye
katkı sağlamıştır.
İmalat sanayi grubu; orman ürünleri sanayisi, ağaç ve mantar ürünleri ile mobilya
sanayisinden oluşmaktadır. Ağaç ve mantar ürünleri ara malı üreten sanayiler arasında
tanımlanırken, mobilya sanayisi ise tüketim malı üreten sanayiler grubunda anılmaktadır [1].
Orman sanayinin alt grubu olarak görülen ahşap sanayisi atölye tipli ve fabrikasyon halinde iki
grup olarak imalat sanayi üretimi içerisinde % 4’lük pay ile yer almaktadır. Ahşap ürünlerine
ilişkin talep her geçen gün artmaktadır. 90’ların sonunda ihracat payı 11 milyon dolar olan
ahşap sektörü 2000’li yılların başında bu payı 180 – 200 milyon dolara kadar çekmiş 2010
yılında 573 milyon doları yakalamış ve 2018 yılında bu rakam 826 milyon dolarlık paya
ulaşmıştır. Türkiye içi piyasada ise 2010 yılının başında ahşap ürüne olan talep 2,6 milyon iken
bu rakam 2017 yılı sonunda 5,5 milyona ulaşmıştır.
Talep; söz konusu ürün grubunda tüketici kesimin bir hizmet veya ürünü belirli bir fiyat
aralığından almaya hazır bulundukları miktardır (Tekin, 1996). Talep tahmini ise; belirlenen
289
zaman periyotları içerisinde gelecekte firmaların söz konusu ürün veya hizmeti için oluşacak
talebin belirlenmesidir (Acar, 1989). Talepler aynı zamanda satış değerlerini de göstereceği için
literatürde ‘talep tahmini’ yerine ‘satış tahmini’ de kullanılmaktadır.
Talep tahmini, firma yönetimi için taktik ve stratejik kararların alınmasında, uzun ve
kısa dönemli hedeflere ulaşmak amacıyla kullanılabilecek araçların başında yer almaktadır.
Talep tahmininin etkinliği şirket fonksiyonlarında optimal kar çizgisine doğru hareket sağlar ve
mevcut hedeflerin gereksinimler ile arasında bulunan çatışmayı minimize eder (Bolt, 1994).
Talep tahmini çalışmaları her uygulamada olduğu gibi belirli kurallar ve yöntemlerin ışığında
yapılır. Fakat seçilen yöntem her ne olursa olsun çıkan sonuçların hepsi kendi özünde doğrudur
ancak bu ifade sonuçların talebi %100 olarak açıklayacağı anlamına gelmemektedir. Söz
konusu ürün için yapılan talep tahmininde ürünün sahip olduğu özellikler ve seçilen yöntem,
tahminlerin tiplerini ve bulunacağı zaman sürelerini etkilemektedir. Ürün için oluşan talepte
dalgalanmalar mevcutsa, yapılacak tahmin en az bir dönemi barındırmak zorundadır. Talep için
uzun dönemde içinde eğilim söz konusu ise, yapılacak tahminin süresi de daha uzun olmalıdır.
Mevsimlere göre talebinde değişiklik meydana gelen ürünlerin talep tahmini yapılırken
değişimlerin sebeplerini görebilmek adına mevsimsel tahmin yöntemleri kullanılmalıdır (Acar,
1999).
Tahmin; incelenen olgunun geçmişinden edinilen bilgiler ışığında, söz konusu olgunun
geleceği hakkında ön görüde bulunmaktır. İncelenen olgunun geçmişi hakkındaki bilgi farklı
yollar ile toplanabilir. Bilgi toplanırken başvurulan yollardan biri ham ve sayısal verileri
matematiksel yöntemler yardımıyla yorumlamaktır. İkincisi ise konu hakkında uzman
görüşlerini ve fikirlerini birleştirerek analizi gerçekleştirmektir (Taşdemir, 2012).
Bu çalışmada; Düzce ilinde ağaç sektöründe üretim faaliyeti gösteren bir işletmede,
gelecek yılların tahmin edilmesinde çoklu doğrusal regresyon analizi ve ağırlıklı hareketli
ortalama yöntemleri kullanılmış ve sonuçlar karşılaştırılmıştır.
Materyal ve Metod
Bu çalışmada Düzce ilinde faaliyet gösteren Antik Atölye firmasına ait 2015 ile 2018 yılları
arasında gerçekleşmiş olan ahşap kullanım verileri kullanılmıştır. Ahşap üretimi kesme,
290
kurutma, depolama, şerit, planya, CNC, kalınlık, montaj ve boyama kalemlerinden
oluşmaktadır.
Tahmin analizinde geçmiş gözlem değerleri kullanılarak, sürecin oluşmasına katkıda
bulunan ilişkiler belirlenir ve bu ilişkilerin geleceği nasıl şekillendireceği tahmin edilmeye
çalışılır.
Tahmin metotları; nitel ve nicel tahmin metotları olmak üzere ikiye ayrılır. Nitel tahmin
yöntemi; tahminde bulunacak kişinin düşüncelerine bağlı, kişiye özgü bir yöntemdir. Nitel
tahmin yöntemleri; matematiksel modellere dayanan yöntemlerdir. Nicel tahminde bulunmak
için iki temel yaklaşım kullanılmaktadır; sebep sonuç ilişkisine ve zaman serileri analizine
dayalı modeller [2]. Nitel tahmin metotlarına; Delphi ve araştırma teknikleri verilebilir. Nicel
tahmin metotları ise zaman serine dayalı yöntemler; Hareketli ortalamalar, Üstel düzeltme,
Arima, Arch, Garch, Tar ve YSA verilebilir. Nicel tahmin metotlarında sebep sonuç ilişkisine
dayalı yöntemler olarak; basit doğrusal regresyon, doğrusal olmayan regresyon, çoklu doğrusal
regresyon ve Yapay sinir ağları (YSA) verilebilir.
Tahminin başarısı önemlidir. Anakütle parametresine ‘yakınlık’ çeşitli ekonometri
tahmin yöntemleri ile bulunmuş tahminlerin örnekteki dağılımların ortalaması ve varyansıyla
ölçülür. İyi bir nokta tahmincisinin özellikleri: Sapmasızlık, minimum varyans (Etkinlik), en
küçük ortalama hata karesi (OHK), tutarlılık ve yeterliliktir [3].
Regresyon analizi, gelecekte oluşacak bilinmeyen olayları mevcut bilinen veriler ile
tahmin edilirken kullanılır. Regresyon analizinde bağımlı değişken ile bağımsız değişken veya
değişkenler arasında olan ilişki ve etkileşmeyi doğrusal eğri yöntemi ile bir tahminde bulunur
(Köse, 2008). Bağımlı ve bağımsız değişkenler arasında olan ilişkiyi gösteren regresyon
denklemi ile bu ilişkide bulunan parametrelerin değerleri bulunabilir. Bağımlı değişken
üzerinde etkisi olan bağımsız değişkenler tahmin edilerek, söz konusu bağımlı değişken
üzerinde ne tür değişimler olacağı, hangi planların uygulanabileceği, bağımsız değişkenlerden
hangisi veya hangilerinin daha önemli ve etkili olduğu ortaya çıkarılabilir. Regresyon analizi
ile bağımsız değişkenlerin bağımlı değişken üzerinde yapmış olduğu etkilerin ne derecede
olduğu, bağımsız değişkenlerde meydana gelecek olan değişimlerin bağımlı değişkeni nasıl
291
etkileyeceği hesaplanabilir (Çağlar, 2007). Bununla beraber istatistiksel yöntemlerle bağımsız
değişkenlerin bağımlı değişkenler üzerinde yaptıkları etkiler gösterilmektedir (Yoldaş, 2006).
Bağımlı bir değişken ile ona etki eden bağımsız değişkenleri inceleyen regresyon
analizinde farklı yöntemler mevcuttur. Çünkü bağımlı ve bağımsız değişkenler arasında
bulunan ilişki eğrisel veya doğrusal olabilmektedir (Aksoy, 2008). Regresyon analizi
değişkenler arasında bulunan ilişkiyi incelerken şu yöntemleri kullanır;
Regresyon analizi, zaman serileri tahmin modellerinin kullanmış olduğu geçmişte
meydana gelen verilerden yola çıkarak gelecekte oluşabilecek olan taleplerin tahmin edilmesi
yöntemini kullanmayıp değişkenler arasında bulunan ilişki üzerinden yola çıkarak tahmin
yapan bir yöntemdir (Üreten, 2013). Regresyon analizi; bağımsız değişkenler (X1, X2 …Xn)
ile bağımlı değişken (Y)’deki değişimi açıklamayı hedefler.
Regresyon modeli iki (ya da daha çok) değişken arasındaki ilişkinin fonksiyonel şeklini
göstermekle kalmaz, değişkenlerden birinin değeri bilindiğinde, diğeri hakkında tahmin
yapılmasını da sağlar.
İki ya da daha çok değişken arasındaki ilişkinin matematiksel bağıntısı “Regresyon
Analizi” ile, ilişkinin yönü ve derecesi ise “Korelasyon Analizi” ile incelenir.
Regresyon analizi bir tahmin (öngörüsel) analizi olup, bağımlı değişkenin bağımsız
değişkenler yardımıyla tahmin edilmesini sağlar. Ayrıca bağımlı değişkeni etkileyen en önemli
bağımsız değişken/değişkenlerin hangisi olduğunu ortaya çıkarır.
Regresyon analizi;
Basit doğrusal regresyon analizi,
Çok değişkenli regresyon analizi,
Doğrusal olmayan regresyon analizi olarak sınıflandırılabilir.
Aralarında ilişki araştırılması istenen değişkenler sayılabilir veya ölçülebilir nitelikte
olabilir. Üzerinde durulan değişkenlerden bağımlı değişken y, bağımsız değişken x ise, y = f(x)
şeklindeki fonksiyona regresyon modeli denir
292
Basit doğrusal regresyon; bağımsız değişken (X) ile bağımlı değişken (Y)’deki değişimi
açıklamayı, bağımsız değişkendeki bir birimlik değişimin bağımlı değişken üzerindeki etkisini
ölçmeyi amaçlar.
Stokastik (Olasılıklı) bir model olan ve anakütledeki ilişkiyi gösteren basit doğrusal
regresyon denklemi aşağıdaki gibi ifade edilir [3]:
𝑦 = 𝛽0 + 𝛽1𝑥 + 𝜀
Burada;
0: Doğrunun y-eksenini kestiği yer ve regresyon sabitidir.
1: Doğrunun eğimi veya regresyon katsayısıdır.
: Rastgele (Tesadüfi-Şans) hata değeridir.
Regresyon denklemi kullanılarak, verilen bir x değeri için y’nin tahmini değeri bulunur;
ancak x ’in büyüklüğü örnek veri setindeki minimum ve maksimum değerler arasında ise daha
iyi tahminler yapılır.
𝑦̂ : y’nin tahmini değeri olarak tanımladığımızda, örneklem için basit regresyon modeli
aşağıdaki gibi gösterilir.
𝑦̂ = 𝑏0 + 𝑏1𝑥
e terimi örneklemden elde edilen hata ve 𝑦̂ tahmini bağımlı değişken değeri olmak üzere,
hata terimi olan 𝑒 = 𝑦 − 𝑦̂ ile gösterilir. Hata terimi her bir gözlem çiftindeki bağımlı değişkene
ilişkin gerçek değer ile modelden tahmin edilen değer arasındaki farktır ve aşağıdaki gibi
gösterilir.
𝜀𝑖 = (𝛽0 + 𝛽1𝑋) − 𝑌𝑖
Regresyon modelinde anakütle hata terimi (epsilon) için genel varsayımlar vardır.
Basit doğrusal regresyon modeli için bu varsayımların sağlanması gereklidir. Bu varsayımlar
aşağıda verilmiştir [3]:
Regresyon modeli parametrelerine ( ‘lara) göre doğrusaldır.
293
Bağımsız değişkenin (X’in) değerleri tekrar eden örneklerde sabittir. Bir başka
deyişle, X’ in stokastik olmadığı varsayılmaktadır.
Regresyon modelinde hataların beklenen değeri sıfırdır.
Regresyon analizinde parametre tahmini “En Küçük Kareler Metodu” (Least Squares
Method) kullanılarak yapılır. Amaç; dağılım grafiğinde (scatter plot) görülen tüm noktalar için,
doğruya uzaklıklarının bulunması ve bunların toplamının minimize edilmesidir.
Regresyon analizini yapmaktaki amaç; örneklem bilgilerinden yararlanarak anakütle
için tutarlı ve güvenilir tahmin değerleri elde etmek olduğundan, gerçek y değerleri ile tahmini
ŷ değerleri arasında farkın olmaması ya da az fark olması beklenir. Bu nedenle β1 ve β2
katsayılarının (𝑦 − ŷ) değerini en küçük yapacak şekilde bulunması gerekir. Bunu sağlayan
yöntemlerden birisi de En Küçük Kareler Yöntemi (EKKY)’dir [3]:
Bu sebepten dolayı maksimum fayda ve verim beklenen en iyi regresyon modelini
kurabilmek adına farkları minimize edecek a ve b değerlerinin bulunması gerekmektedir.
Basit doğrusal regresyondan farklı olarak çoklu doğrusal regresyon bağımlı değişkenin
birden fazla bağımsız değişken ile ilişkili olduğu durumlarda kullanılmaktadır (Yılmaz ve Top,
2009). Çoklu regresyon analizinde birden fazla bağımsız değişkenin bağımlı değişken ile olan
ilişkileri incelenmektedir. Regresyon analizine dahil edilen bağımsız değişkenin bağımlı
değişkeni ne düzeyde açıklayabildiği çoklu regresyon modellerinde önemli bir noktadır.
Regresyon analizinde bağımlı değişken üzerinde etkisi olan bağımsız değişkenlerin katsayıları
etki düzeylerini göstermediğinden korelasyon analizi yardımı ile bağımsız değişkenlerin etki
düzeyleri bulunur (Karaca, 2015). Çoklu doğrusal regresyon yöntemine ait matematiksel
denklem ‘Denklem 1’de gösterilmiştir.
Denklem 1 : 𝑌𝑖 = 𝑎 + 𝑏1𝑋1 + 𝑏2𝑋2 + ⋯ + 𝑏𝑛𝑋𝑛
Burada 𝑌𝑖 hesaplanmak istenen bağımlı değişkeni, a regresyon doğrusuna ait başlangıç
değerini, b regresyon doğrusuna ait olan eğimi ve 𝑋𝑖 bağımsız değişkenleri ifade etmektedir
(Aksoy, 2008). ’Denklem1’de b katsayıları bağımsız değişkenlerde oluşacak değişimin bağımlı
değişken üzerinde bıraktığı etkiyi göstermektedir. Çoklu doğrusal regresyon fonksiyonuna
ulaşma yolunda basit doğrusal regresyon fonksiyonunda olduğu gibi en küçük kareler
yönteminden yararlanılır (Yoldaş, 2006).
294
Doğru denklemleri değişkenler arasında bulunan ilişkiyi ifade etmekte her zaman yeterli
olmayabilir. Bu gibi durumlarda eğri denklemleri kullanılır. Bağımlı değişken ile bağımsız
değişkenler arasındaki ilişkiyi bulmak için kullanılan eğri denklemlerinde en kolay
uygulanabilecek işlem verileri grafik olarak çizmektir. Veriler sonucunda ulaşılan eğriler üssel,
hiperbolik ya da parabolik olabilirler. Modellerin belirlenmesinden sonra değişkenler arasında
bulunan ilişkiyi en verimli ve en sağlam açıklayan model uygun olarak seçilmelidir (Özsoy,
2006).
Bağımlı değişken Y ve bağımsız değişken X olarak tanımlandığında y = f(x) şeklindeki
fonksiyona regresyon modeli denir. f(x) fonksiyonu aşağıda farklı şekiller alabilir (Ersöz,
Ersöz, 2019):
Doğrusal: 𝑦 = 𝑎𝑥 + 𝑏
Parabolik: 𝑦 = 𝑎𝑥 2 + 𝑏
Üstsel: 𝑦 = 𝑎𝑏 𝑥 , 𝑦 = 𝑎𝑒 𝑥
Geometrik: 𝑦 = 𝑎𝑥 𝑏 → log 𝑦 = 𝑏 log(𝑎𝑥)
Hiperbolik: 𝑦 = (𝑎𝑥 + 𝑏) −1
Ağırlıklı hareketli ortalama yöntemi genel olarak basit hareketli ortalama yöntemine
benzemektedir. Hesaplanmasında son dönemlere ait verilerin toplanarak aritmetik ortalaması
alınan ve eşit değerde ağırlık atanan basit hareketli ortalama yönteminden farklı olarak bu
yöntemde hesaplamaya dahil edilen dönemlere farklı farklı ağırlıklar atanabilmektedir (Özer,
2009). Bir diğer ifadeyle, yapılacak tahmine etki düzeyi düşünüldüğünde daha fazla etkiye
sahip olacak döneme diğer dönemlere nazaran daha büyük ağırlık, daha az etkiye sahip olacak
döneme diğer dönemlere nazaran daha küçük ağırlık atanmaktadır. Bu yöntemde tahmin,
belirlenen dönemlerin talepleri atanan ağırlıklar ile çarpılır. Elde edilen çarpım sonuçları
toplanır ve daha sonra ağırlıkların toplamına bölünür. Tahmin sonucu elde edilir (Taşdemir,
2012).
Ağırlıklı hareketli ortalama yönteminde dönemlere atanan ağırlık değerleri 0 ile 1
arasında değerler almaktadır ve ağırlık toplamları 1’e eşit olur (İlhan, 2015). Üç aylık dönem
baz alınarak yapılan ağırlıklı hareketli ortalama yönteminde son döneme ait ağırlık katsayısı
295
0,5, son dönemden bir önceki döneme ait ağırlık katsayısı 0,3 ve en uzak döneme ait ağırlık
katsayısı 0,2 olarak örneklendirilebilir (Malhotra, 2014). Ağırlıklı hareketli ortalama
yönteminde tahmin edilen döneme en yakın dönemin ağırlık katsayısı büyük verilerek hareketli
ortalama yönteminde her dönemin eşit ağırlığa sahip olması problemi çözüme ulaştırılır
(Demirbaş, 2011). Ağırlıklı hareketli ortalama yöntemine ait matematiksel formül ‘Denklem 2’
de gösterilmiştir.
Denklem 2 : 𝑆𝑡 = 𝑊𝑡−1𝐴𝑡−1+𝑊𝑡−2𝐴𝑡−2+⋯+𝑊𝑡−𝑛𝐴𝑡−𝑛 ∑𝑊
Burada 𝑆𝑡 tahmin değeri olarak ifade edilirken, hareketli ortalamaya dahil edilen dönem
sayısı n ile, dönemler içerisinde gerçekleşen değerler A ile ve söz konusu dönemler için
belirlenen ağırlık değerleri W ile ifade edilmektedir (Yüksel, 106).
Hareketli ortalama yönteminden farklı olarak ağrılıkların hesaba dahil edilerek istenen
veriler adına ortalama içerisindeki pay artırılmaktadır. Tahminlerin son dönemlerde
gerçekleşen değişikliklere daha hızlı geri dönüş, cevap vermesi, son dönemlere verilen
ağırlıkların büyük olmasına bağlıdır. Bu aşamada en önemli nokta dönemlere verilen
ağırlıkların belirlenmesidir. Ağırlıklar deneme yanılma yönteminden faydalanılarak
belirlenmektedir (Yılmaz ve Top. 2009).
Bulgular
Bu çalışmada, Düzce ilinde faaliyet gösteren Antik Atölye firması adına gelecekte oluşacak
ağaç sarfiyatını hesaplamak için 2015Q1 ve 2018Q12 arasındaki dönemlere ait veriler
kullanılarak nüfus artış oranı, kereste fiyatı ve TÜFE değişkenlerine ait aynı dönemdeki
verilerin söz konusu atölye adına ağaç sarfiyatına olan etkisi araştırılmıştır. Çalışmada çoklu
doğrusal regresyon ve ağırlıklı hareketli ortalama yöntemleri kullanılmış ve SPSS Statistics22
paket programından yararlanarak analizler yapılmıştır. Çalışma sonucunda her iki yöntem adına
bulunan sonuçlar kıyaslanmış ve yorumlanmıştır.
Tek bağımsız değişken olduğunda bakılan R değeri ve çalışmada kullanılan gibi birden
fazla bağımsız değişken olduğunda değerlendirilen ARS değeri; bağımsız değişkenlerin
bağımlı değişkenin varyansının yüzde kaçını açıkladığını söylemektedir. Yani nüfus artış oranı,
kereste fiyatı ve Tüfe Antik Atölye adına kullanılan kereste miktarının %67,2 sini
açıklamaktadır.
296
Kurulan modelin anlamlılığı Anova ile araştırılmış ve modelin anlamlı olduğu
görülmüştür. Yani bağımsız değişkenlerin en birinin bağımlı değişken üzerinde anlamlı bir
etkiye sahip olup olmadığını göstermektedir.
Modelde değişkenler arasında çoklu bağlantı problemi bulunmamaktadır. Bütün haliyle
model %5 anlamlılık düzeyinde anlamlıdır. Bağımsız değişkenlerin katsayıları tek tek
incelendiğinde %5 anlamlılık düzeyinde sabit terim katsayısı 0,000 ile, nüfus artış oranı
katsayısı 0,001 ile, kereste fiyatı katsayısı 0,000 ile, α = 0,05 değerinden küçük olup anlamlı
oldukları anlaşılmaktadır. Fakat bununla beraber Tüfe katsayısı 0,911 ile anlamsız çıkmıştır.
Ek olarak burada nüfus artış oranındaki değişim ile bağımlı değişken arasında ters yönlü ilişki
vardır. R2 değeri %67,2 bulunmuş ve bağımsız değişkenlerin bağımlı değişkeni açıklama oranı
yeterli görülmüştür.
Bu bağlamda regresyon denklemi;
𝑌 = 13,721 − 3,626(𝑁ü𝑓𝑢𝑠 𝐴𝑟. ) + 0,014(𝐾𝑒𝑟𝑒𝑠. 𝑓𝑖𝑦. ) − 0,04(𝑇ü𝑓𝑒) şeklinde
tanımlanabilir. Bu fonksiyonda yer alan sabit değer, nüfus artış oranı ve kereste fiyatı söz
konusu atölye için önerilen üretim tahmin fonksiyonunda yer almalıdır. Tüfe değişkeni %5
anlamlılık düzeyinde anlamsız olduğundan önerilen üretim tahmin fonksiyonunda yer almasına
gerek yoktur.
Bu çalışmada ayrıca talep tahmini için zaman serisi yöntemlerinden ağırlıklı hareketli
ortalama yöntemi kullanılmıştır. Çalışmada kullanılan dönemler adına hesaplama yapılırken üç
aylık periyotlar oluşturulmuştur.
𝑆𝑡 = 𝑊𝑡−1𝐴𝑡−1 + 𝑊𝑡−2𝐴𝑡−2 + ⋯ + 𝑊𝑡−𝑛𝐴𝑡−𝑛 ∑ 𝑊
Formülü doğrultusunda 48 dönem için farklı tahmin sonuçları elde edilmiştir. Ağırlıklı
hareketli ortalama yönteminde her dönem için birbirinden farklı ağırlıklar uygulanmaktadır.
Son döneme ait verilerin tahmin yapılırken etkisinin daha büyük olacağı kanası mevcutsa o
dönemin ağırlığı daha fazla olur. Bu çalışmada tahmin yapılırken son döneme ait ağırlık
katsayısı 0,5, bir önceki döneme ait ağırlık katsayısı 0,3 ve yine ondan bir önceki döneme ait
ağırlık katsayısı 0,2 olarak alınmıştır. Çalışmada tahmin edilen değerler ile gerçekleşen değerler
kıyaslandığında değerlerin birbirine yakın olduğu görülmektedir. Bu durumda bu tahmin
yöntemi gelecek yıllar adına yapılacak üretim planlaması için uygun olacaktır. Gerçekleşen
297
değerler ile tahmin edilen değerlerin ortalaması alındığında %98’lik bir oran çıktığı ve tahmin
yönteminin gayet başarılı olduğu görülmektedir.
Sonuç ve Tartışma
Hızla gelişen teknoloji çağında işletmeler müşterilerine maksimum verimli ve minimum
maliyetli hizmetler sunma eğilimindedir. İşletmelerin bu söz konusu kısıtları en iyi şekilde
yerine getirebilmesi için ileride nelerle karşılaşacaklarını bilmeleri ya da tahmin etmeleri
gerekmektedir. Bu bağlamda işletmeler için talep tahmini çalışmaları oldukça önemli bir
noktaya sahiptir.
Bu çalışmada Düzce ilinde faaliyet gösteren Antik Atölye firması adına 2015Q1 ile
2018Q12 dönemleri arasında gerçekleşen veriler ışığında gelecek dönemler adına tahmin
yapılabilmesi için bir tahmin denklemi oluşturulmaya çalışılmış ve kullanılan yöntemler adına
çıkan sonuçlar kıyaslanarak hangi yöntemin daha etkili olduğu ortaya konmak istenmiştir. Bu
amaçla çalışmada çoklu doğrusal regresyon analizi yöntemi ve ağırlıklı hareketli ortalama
yöntemleri kullanılmıştır.
Regresyon analizi sonucunda söz konusu firma için üretim tahmini formülü 𝑌 = 𝑁
−0,362 + 𝐾 0,566 + 𝑇 −0,010 şeklinde bulunmuştur. Üretim tahmin fonksiyonunda Antik
Atölye firması adına nüfus artış oranının ters yönlü anlamlı bir ilişkisi olduğu, kereste fiyatında
doğru orantılı anlamlı bir ilişki olduğu görülmüştür. Tüfe değerinin firmanın üretim tahmini
adına anlamsız olduğu saptanmıştır.
Ağırlıklı hareketli ortalama yönteminde ise firmadan alınan veriler ile üç aylık
periyotlara üzerinden tahminlerde bulunulmuş ve gerçekleşen değerler ile kıyaslanmıştır.
Çalışmada kullanılan iki yöntem karşılaştırıldığında ağaç endüstrisinde faaliyet gösteren
firma için ağırlıklı hareketli ortalama yönteminin tahmin değerleri regresyon analizi yönteminin
tahmin değerlerinden daha iyi sonuç verdiği gerçekleşen değerler ile tahmin edilen değerlerin
farkından ortaya konmuştur.
Literatürde talep tahmini için birçok yöntem bulunmaktadır. Farklı sektörlerde farklı
yöntemler daha iyi sonuç vermektedir. Bu çalışmada ağırlıklı hareketli ortalama yöntemi
298
gerçekleşen değerlere daha yakın sonuçlar vermiştir. Bununla beraber yapay sinir ağları
yöntemi ile tahminleme yapılması daha hassas sonuçlar elde etmek için denenebilir.
KAYNAKÇA
[1] H. Cındık, H., Akyüz, C. ve Serin, “Orman Ürünleri Sanayinde Küçük ve Orta Öçekli
İşletmeler,” in 1. Ulusal Mobilya Kongresi Bildiri Kitabı, 1997, pp. 176–189.
[2] J. S. Montgomery, D.C., Johnson, L.A. and Gardiner, Forecasting and Time Series Analysis.
NY: McGraw Hill, 1990.
[3] E. T. Ersöz F., İstatistik – I. Ankara: Seçkin Yayınevi, 2019.
[4] F. Ersöz; IBM SPSS ile İstatistiksel Veri Analizi, Seçkin Kitabevi, 2019
299
Deep Learning Based Abnormality Detection Application in Enterprise Network Traffic
Emrullah ERGİNAY, M. Ali AKÇAYOL
Department of Computer Engineering, Gazi University, TURKEY
Abstract: In this paper, a deep learning model has been developed to detect whether
malware/spyware leaks data to command and control servers and a new dataset has been
obtained from real-time environment for test of the model. In addition, effect of the size of the
data set and hyperparameters such as the number of layers of the deep neural network on the
success rate have been investigated. In this study, real-time data for harmful and normal İnternet
traffic have been obtained in the application layer and 100 features have been selected. The
developed deep learning model has been applied to 16,000 sample obtained from real-time
Internet traffic. From the experimental results, accuracy rates of 90% to 94% were obtained
with various number of samples and various number of layers in the deep learning model. It has
been seen from the experimental results that increase the number of samples increases the
accuracy rate. As well as, it has been seen that as increase the number of layers in the deep
neural network the accuracy rate increased first, further increase the hidden layers did not affect
the success rate. In this study, more distinctive and important features have been investigated
than others in the literature and the results have been tested.
Keywords: Anomaly Detection, Deep Learning, Corporate Networks, Malware
REFERENCES
[1] D. Zhao, I. Traore, B. Sayed, W. Lu, S. Saad, A. Ghorbani and D. Garant, “Botnet detection
based on traffic behavior analysis and flow intervals”, Computers & Security 39, 2-16, April 2013.
[2] M. Ichino, K. Kawamoto, T. Iwano, M. Hatada and H. Yoshiura, “Evaluating Header
Information Features for Malware Infection Detection”, Journal of Information Processing, Vol.23,
No.5, 603-612, September 2015.
[3] S. Kumar, A. Viinikainen and T. Hamalainen, “Machine Learning Classification Model For
Network Based Intrusion Detection System”, The 11th International Conference for Internet
Technology and Secured Transactions (ICITST-2016), 242 – 249, December 2016.
300
[4] K. Singh, S. Guntuku, A. Thakur and C. Hota, “Big Data Analytics framework for Peer-toPeer Botnet detection using Random Forests”, Information Sciences 278, 488-497, March 2014.
[5] A. Boukhtouta, S. Mokhov, N. Lakhdari, M. Debbabi and J. Paquet, “Network malware
classification comparsion using DPI and flow packet headers”, Journal of Computer Virology and
Hacking Techniques, 69-100, July 2015.
[6] M. Mimura, Y. Otsubo, T. Hidehiko and T. Hidema, “A Practical Experiment of the HTTPBased RAT Detection Method in Proxy Server Logs”, 12th Asia Joint Conference on Information
Security, 31-37, August 2017.
[7] L. Boero, M. Marchese and S. Zappatore, “Support Vector Machine meets Software Defined
Networking in IDS domain”, 29th International Teletraffic Congress (ITC 29), 25-30, September 2017.
[8] A. Verma and V. Ranga, “Statistical analysis of CIDDS-001 dataset for Network Intrusion
Detection Systems using Distance-based Machine Learning”, Procedia Computer Science 125, 709-716,
2018.
[9] Z. Chen, Q. Yan, H. Han, S. Wang, L. Peng, L. Wang and B. Yang, “Machine learning based
mobile malware detection using highly imbalanced network traffic”, Information Sciences, 346-364,
April 2018.
[10] A. Lashkari, A. Kadir, H. Gonzalez, K. Mbah and A. Ghorbani, “Towards a Network-Based
Framework for Android Malware Detection and Characterization”, 15th International Conference on
Privacy, Security and Trust, August 2017.
[11] V. Kant, M. Singh and N. Ojha, “An Efficient Flow based Botnet Classification using
Convolution Neural Network”, International Conference on Intelligent Computing and Control Systems
(ICICCS 2017), 941-946, June 2017.
[12] T. Shibahara, T. Yagi, M. Akiyama, D. Chiba and T., “Efficient Dynamic Malware
Analysis Based on Network Behavior Using Deep Learning”, Global Communications Conference
(GLOBECOM), December 2016.
[13] D. Y. Kim and H. Y. Song, “Method of Predicting Human Mobility Patterns Using Deep
learning”, Neurocomputing 280, 56-64, March 2018.
[14] S. Lin, K. Ying, C. Lee and Z. Lee, “An İntelligent Algorithm With Feature Selection And
Decision Rules Applied To Anomaly İntrusion Detection”, Applied Soft Computing 12, 3285-3290,
October 2012.
301
[15] K. P. Ferentinos, “Deep learning models for plant disease detection and diagnosis”,
Computers and Electronics in Agriculture 145, 311-318, February 2018.
[16] Ç. Kaya and O. Yıldız, “Makine Öğrenmesi Teknikleriyle Saldırı Tespiti: Karşılaştırmalı
Analiz”, Marmara Fen Bilimleri Dergisi 2014, 3:89-104, March 2015.
[17] S. B. Kotsiantis, “Supervised Machine Learning: A Review of Classification Techniques”,
Informatica 31 (2007), 249-268, July 2007.
[18] A. Kamilaris and F. Prenafeta-Boldu, “Deep Learning in Agriculture: A Survey”,
Computers and Electronics in Agriculture 147, 70-90, April 2018.
[19] A. A. Diro and N. Chilamkurti, “Distributed Attack Detection Scheme Using Deep
Learning Approach For Internet of Things”, Future Generation Computer Systems 82, 761-768, May
2018.
[20] J. Gu, Z. Wang, J. Kuen, L. Ma, A. Shahroudy, B. Shuai, T. Liu, X. Wang, G. Wang, J. Cai
and T. Chen, “Recent advances in convolutional neural networks”, Pattern Recognition 77, 354-377,
October 2018.
[21] W. Wang, M. Zhu, X. Zeng, X. Ye and Y. Sheng, “Malware Traffic Classification Using
Convolutional Neural Network for Representation Learning”, Information Networking (ICOIN),
January 2017.
[22] T. Tang, L. Mhamdi, D. McLernon, S. A. R. Zaidi and M. Ghogho, “Deep Learning
Approach for Network Intrusion Detection in Software Defined Networking”, Wireless Networks and
Mobile Communications (WINCOM), October 2016.
[23] H. Chougrad, H. Zouaki and O. Alheyane, “Deep Convolutional Neural Networks for breast
cancer screening”, Computer Methods and Programs in Biomedicine 157, 19-30, April 2018.
302
Lean Production in Iron and Steel Production Line
Taner ERSÖZ1, Filiz ERSÖZ1, Karanfil SARIZ2
2
2
Department of Actuary and Risk Management,
Department of Industrial Engineering, Karabük University, TURKEY
Abstract: Rapidly developing technology in the world endangers many companies we are in
today. The companies are now obliged to improve themselves in order to compete in both
domestic and international markets. In this sense, concepts such as quality, cost and customer
satisfaction are becoming more important. Lean production is a production method that is not
outdated at this point. In the 1950s, the lean production which was laid the foundation with the
Toyota Production Company, is still up to date. Lean production, which adopts zero error and
zero inventory principle, is applied in both the manufacturing sector and the service sector.
Today, it is still very important in terms of the continuity of lean production which is developed
and integrated with new technologies. Lean production is a gateway to the world from the
Japanese people. In this thesis,” Lean Manufacturing Application in Iron-Steel Production Line
Hatt is tried to be explained. Lean production and techniques are examined in detail. Then, in
light of this information, the results obtained in iron and steel industry were interpreted.
Keywords: Lean, Just in time, Profit, Cost, Waste, Simulation
REFERENCES
[1] Arslan, S., “Yalın Üretim ve Man Türkiye A.Ş.’ de Örnek Bir Yalın Üretim Enstitüsü,
Ankara, 3-43 (2008).
[2] Atmaca, M., Terzi, S., “Yalın Üretim Sistemi Açısından Değer Akışı Maliyetlemesinin
İncelenmesi”, Fakülte Dergisi, 16, 449-466 (2011).
[3] Michalska, J., Szewieczek, D., “The 5S Methodology As a Tool For İmproving The
Organisation”, Journal of Achievements in Materials and Manufacturing Engineering, 2: 211214(2007).
303
[4] Ohno, T., “Toyota Ruhu”, Hakan Feyyat/ Canan Feyyat, İstiklal Caddesin Han Geçidi
Sokak, Galatasaray/İstanbul,39-197 (2017).
[5] Sarı, Ö., Seri, V. ve Yazgan, H., “Toyota Üretim Sisteminin Özellikleri”, SAÜ Fen Bilimleri
Enstitüsü Dergisi, 2, 129-134 (1998).
[6] Şeker, A., “Yalın Üretim Sisteminde Kanban, Tek Parça Akışı ve U Tipi Yerleştirme
Sistemleri”, The Journal of Academic Social Science Studies, 449-470 (2016)
[7] Türkan, Ö., “Üretimde Yalın Dönüşümün Temel Performans Kriterleri”, BAÜ Fen Bilimleri
Enstitüsü Dergisi, 12, 28-41 (2010).
[8] Yılmaz, E., “Siparişe Göre Üretim Yapan Sistemlerde Yalın Üretim Uygulamaları”, Yüksek
Lisans Tezi, İstanbul Teknik Üniversitesi Fen Bilimleri Enstitüsü, İstanbul, 11-57 (2012).
[9] İpek, F., “Siparişe Göre Üretimde Malzeme Hazırlama ve Üretim Hattının Beslenmesinin
İyileştirilmesi: Yalın Üretim Uygulaması”, Yüksek Lisans Tezi, Gazi Üniversitesi Fen Bilimleri
Enstitüsü, Ankara, 3-28 (2015).
[9] İraz, R., Zerenler, M., “Japon yönetim Anlayışı ve Şirket Ağları (KEIRETSU) Analizi,
Balıkesir Üniversitesi Sosyal Bilimler Dergisi, 758- 773.
[10] Jones, D., Womack, J., “Yalın Düşünceler”, Oygur Yamak, Osmangazi Mahallesi/İstanbul,
21-387 (2016).
[11] Jones, D., Womack, J., “Yalın Çözümler”, Saadet Özkal, Akçaburgaz Mahallesi
Esenyurt/İstanbul, 15-143(2015).
[12] Karaboğa, K., Zerenler, M., “Müşteri Memnuniyetinin Sağlanmasında Hataların
Önlenmesine Yönelik Üretim Odaklı Bir Bakış Açısı: PokaYoke Sistemleri”, Sosyal Bilimler Enstitüsü
Dergisi, 263-275 (2014).
[13] Kaymakçı, Ö., “Bir PTT Şubesinde Yalın Üretim-5S Uygulaması”, Yüksek Lisans Tezi,
Sakarya Üniversitesi Fen Bilimleri Enstitüsü, Sakarya, 4-57 (2012).
[14] Kılıç, A., “Otomotiv Yan Sanayinde Yalın Üretim Uygulaması”, Yüksek Lisans Tezi,
İstanbul Üniversitesi Fen Bilimleri Enstitüsü, İstanbul, 3-51 (2016) Rother M., Shook J., “Görmeyi
Öğrenmek”, Ayşe Soydan, Merkez Mah. Sefir Sokak Sarıyer/İstanbul,1-102 (1999).
304
A Road Map for The Big Data Implementation in the Judicial System
Zeynep ÇAĞLAR
Sakarya University, TURKEY
Abstract: This research involves Big Data and its definition. Also, it covers general
visualization of the problems in Turkish legal system and deal with a real case to create a
methodology for solving the problems in filing an indictment process with the help of Big Data.
The goal is to show that how can the finishing time of the process of filing indictments
significantly reducing when Big Data technology integrated to the process. This has been done
by creating a decision support system that can apply to all the cases in the phase of linking
evidences and filling the indictment. This research will provide valuable information regarding
the methodology of applying Big Data and address the role of Big Data in solving problems of
ineffective and complex organizations.
Keywords: Big Data Road Map, Indictment, Methodology.
REFERENCES
[1] Laney, D. (2001) 3D Data Management: Controlling Data Volume, Velocity and
Variety. Working paper, Application Delivery Strategies, Meta Group AG.
[2] Lämmel, R. (2008) Google's Map Reduce Programming Model, Revisited. Science
of Computer Programming. 70: 1–30
[3] Grover, P. and Kar, A. (2017). Global Journal of Flexible Systems Management.
Mumbai: Springer India, pp.203-229.
305
Supplier Selection Using an Intuitionistic Fuzzy Evaluation System
Ayşenur AKIN1, M. Bahar BAŞKIR1, Hamza GAMGAM2
1
2
Department of Statistics, Bartın University,
Department of Statistics, Gazi University, TURKEY
Abstract: Selection and evaluation problems have uncertainties due to concept and its
perception discrepancies. Fuzzy set theory is one of the widely used methodology to cope with
these uncertainties. There is a growing interest in evaluations using intuitionistic fuzzy sets.
Differently from fuzzy sets, intuitionistic sets include both belonging degrees and nonbelonging
degrees. Thus, the evaluation using intuitionistic fuzzy sets gives more realistic results. In this
study, an intuitionistic fuzzy set-based evaluation system is proposed for supplier selection
problem of a construction company. This system has qualitative- and quantitative evaluation
parts. As qualitative part, decision makers of the company evaluate suppliers by the supplier
selection criteria: i) quality, ii) price, iii) delivery, iv) productivity, v) service, vi) flexibility.
The quantitative part includes supplier scores calculated through the current evaluation system
of the company. A supplier-evaluation database was created by the abovementioned parts. The
database was structured by α-cut representation of the evaluations. The calculations using
intuitionistic fuzzy sets were done for this database, which was occurred by lower and upper
bounds. After defuzzifying the database, suppliers were classified using the well-known fuzzy
clustering algorithm, fuzzy c-means. The classification using fuzzy clustering algorithm has
95.0% accuracy.
Keywords: Intuitionistic Fuzzy Sets, Supplier Selection, Fuzzy Clustering, α-Cuts, Accuracy.
306
Environmental Impact Assessment of Two Alternative Wastewater Neutralization
Chemicals in Textile Industry Wastewater Treatment Plant
Fatma Şener FİDAN1, Emel Kızılkaya AYDOĞAN1, Niğmet UZAL2
1
2
Dept. of Industrial Engineering, Erciyes University,
Dept. of Civil Engineering, Abdullah Gül University, TURKEY
Abstract: The wastewater treatment in the textile industry is of special importance due to the
intensive use of chemicals and dyes. However, wastewater treatment plants have impacts on
environmental and these environmental impacts should also be evaluated and minimized.
Neutralization is one of the processes with the huge chemical consumption in the wastewater
treatment plant. Therefore, the chemical alternatives used in the neutralization process should
be compared in terms of their environmental impacts. In this study, the performances of carbon
dioxide and sulfuric acid as two alternative chemicals used in the neutralization process applied
in a textile factory wastewater treatment plant are compared using the life cycle approach. The
neutralization process using carbon dioxide yielded better results in the categories of abiotic
depletion, fossil fuels, ozone layer depletion (ODP), fresh aquatic ecotoxicity, marine aquatic
ecotoxicity, terrestrial ecotoxicity, photochemical oxidation, acidification, and eutrophication.
Keywords: Wastewater treatment, Sustainability, Life Cycle Assessment, Textile Industry,
Neutralization.
REFERENCE
[1] WCED, 1987. World Commission on Environment and Development, Our Common Future.
Oxford University Press, New York.
[2] European Commission. Integrated pollution prevention and control (IPPC). Best Available
Techniques
(BAT)
Reference
Document
http://eippcb.jrc.ec.europa.eu/reference/
307
for
Waste
Treatment,
2018,
[3] Baumann, H., Tillman, A.M., 2004. The Hitch Hiker's Guide To LCA: An Orientation in
Life. Cycle Assessment Methodology and Application. Studentlitteratur, Lund.
[4] ISO, 2006a. ISO 14040 Environmental Management -- Life Cycle Assessment -- Principles
and Framework, International Organization for Standardization, Geneva.
[5] ISO, 2006b. ISO 14044: Environmental Management -- Life Cycle Assessment -Requirements and Guidelines, International Organization for Standardization Internatio, Geneva
[6] Raghuvanshi, S., Bhakar, V., Sowmya, C., Sangwan, K.S., 2017. Wastewatertreatment Plant
Life Cycle Assessment: Treatment Process to Reuse of Water. The 24th Cirp Conference on Life Cycle
Engineering, Procedia Cirp 61, 761 – 766.
[7] Yildirim, M., Topkaya, B., 2012. Assessing Environmental Impacts of Wastewater
Treatment Alternatives for Small-Scale Communities. Clean E Soil, Air, Water 40 (2), 171e178.
[8] Venkatesh, G., Brattebø, H., 2011. Environmental Impact Analysis of Chemicals and Energy
Consumption in Wastewater Treatment Plants: Case Study of Oslo, Norway. Water Science and
Technology 63 (5), 1018e1031.
[9] Tillman, A., Svingby, M. And Lundstrom, H., Life Cycle Assessment Of Municipal
Wastewatersystems, The International Journal Of Life Cycle Assessment, Vol. 3, No. 3, Pp. 145-157,
1998, Http://Dx.Doi.Org/10.1007/Bf02978823
[10] Vidal, N., Poch, M., Martí, E., Rodríguez-Roda, I., 2002. Evaluation of The Environmental
Implications to Include Structural Changes in A Wastewater Treatment Plant. Journal of Chemical
Technology & Biotechnology, Vol. 77, No. 11, Pp 1206-1211, Http://Dx.Doi.Org/10.1002/Jctb.674
[11] Remy, C., Jekel, M., 2008. Sustainable Wastewater Management: Life Cycle Assessment
of Conventional and Source-Separating Urban Sanitation Systems. Water Science and Technology 58
(8), 1555e1562.
[12] Pasqualino, J. C., Meneses, M., Castells, F., 2011. Life Cycle Assessment of Urban
Wastewater Reclamation and Reuse Alternatives. J. Ind. Ecol., Vol. 15, No. 1, Pp 49-63,
Http://Dx.Doi.Org/10.1111/J.1530-9290.2010.00293.X
[13] Y. Li, X. Luo, X. Huang, D. Wang, W. Zhang, 2013. Life Cycle Assessment of a
Municipal Wastewater Treatment Plant: A Case Study in Suzhou, China. J. Clean. Prod., Vol. 57,
221–227.
[14] Opher, T., Friedler, E., 2016. Comparative LCA of Decentralized Wastewater Treatment
Alternatives for Non-Potable Urban Reuse. J. Environ. Manag. 182, 464–476.
308
[15] Alfonsín, C., Hospido, A., Omil, F., Moreira, M.T., Feijoo, G., 2014. Ppcps In
Wastewater Update and Calculation of Characterization Factors for Their Inclusion in LCA Studies.J.
Clean. Prod. 83, 245–255.
[16] Guest, J.S., Skerlos, S.J., Barnard, J.L., Beck, M.B., Daigger, G.T., Hilger, H., Jackson,
S.J., Karvazy, K., Kelly, L., Macpherson, L., Mihelcic, J.R., Pramanik, A., Raskin, L., Van
Loosdrecht, M.C.M., Yeh, D., Love, N.G., 2009. A New Planning and Design Paradigm to Achieve
Sustainable Resource Recovery from Wastewater. Environmental Science and Technology 43 (16),
6121e6125.
[17] Ecoinvent Centre, 2013. Database Ecoinvent Data V3.0, Swiss Centre for Life Cycle
Inventories. URL Www.Eco-Invent.Org
[18] Pre Sustainability, 2017. Pré Consultant, Simapro 8.5.2.0, Https://Simapro.Com/, Geneva.
[19] Guinée J. B., Gorrée M., Heijungs R., Huppes G., Kleijn R., De Koning A., Van Oers L.,
Wegener Sleeswijk A., Suh S., Udo De Haes H. A., De Bruijn H., Van Duin R., Huijbregts M. A. J.,
Lindeijer E., Roorda A. A. H. And Weidema B. P., 2001a. Life Cycle Assessment; An Operational
Guide to The ISO Standards; Parts 1 And 2. Ministry of Housing, Spatial Planning and Environment
(VROM) And Centre of Environmental Science (CML). Den Haag And Leiden, The Netherlands,
Retrieved From: Http://Www.Leidenuniv.Nl/Cml/Ssp/Projects/Lca2/Lca2.Html
[20] Guinée J. B., Gorrée M., Heijungs R., Huppes G., Kleijn R., De Koning A., Van Oers L.,
Wegener Sleeswijk A., Suh S., Udo De Haes H. A., De Bruijn H., Van Duin R., Huijbregts M. A. J.,
Lindeijer E., Roorda A. A. H. And Weidema B. P., 2001b. Life Cycle Assessment; An Operational
Guide to The ISO Standards; Part 3: Scientific Background. Ministry of Housing, Spatial Planning and
Environment (VROM) And Centre of Environmental Science (CML). Den Haag And Leiden, The
Netherlands, Retrieved From: Http://Www.Leidenuniv.Nl/Cml/Ssp/Projects/Lca2/Lca2.Html.
309
Empati Çıkar Yöntemiyle Akademik Çalışma Gruplarının İncelenmesi
Ömer Faruk ACAR1, Burhan SELÇUK2
1
Yüksek İhtisas University, 2Karabük University, TURKEY
Abstract: Bu çalışmada akademisyenlerin yayınları kullanılarak yayından yazara birliktelik
analizi yapılmış ve akademik çalışma grupları oluşturulmaya çalışılmıştır. Yayınların başlık,
anahtar kelime, özetlerinde geçen kelimeler morfolojik olarak köklerine ayrılmış, metinsel
temizleme yapılmış ve Empati-Çıkar yöntemi kullanılarak benzerlik/yakınlık katsayıları
hesaplanmıştır. Oluşan benzerlik/yakınlık matrisi kullanılarak yayında katkısı bulunan yazarlar
yakınlıklarına göre çalışma gruplarına dahil edilmeye çalışılmıştır.
Keywords: Semantik Ağ, Graf Yapısı, Veri Madenciliği, Türkçe Morfoloji, Birliktelik Analizi,
Karmaşık Ağ, Empati - Çıkar.
310
Application and Comparison of Biclustering Methods in Detecting Crime Regions
Nazan SARI, Sümeyye Gizem ÇAKAR, Olcay EYDEMİR, İbrahim ÇİL
Department of Industrial Engineering, Sakarya University, TURKEY
Abstract: Crime analysis has importance for the detection of crime regions, the prediction of
crimes before processing and the security forces to take necessary measures. By using
biclustering methods to detect crime regions, simultaneous clustering of the types of crimes and
regions where crime is committed to producing more comprehensive results than traditional
clustering methods. In this study, CC and Xmotif algorithms of biclustering methods were
applied to the real data set in order to detect the crime regions. “Crimes in Boston” data set was
used in real data set application. In order to measure the efficiency of the biclusters, the
performance of the algorithms was compared with Chia and Karuturi bicluster score (CCPS).
The results were obtained by using Matlab functions and it was observed that results of the CC
algorithm were better compared to Xmotif algorithm.
Keywords: Biclustering, CC, Xmotif, Crime Data.
REFERENCES
[1] G. Eason, B. Noble, and I. N. Sneddon, “On Certain Integrals of Lipschitz-Hankel
Type Involving Products of Bessel Functions,” Phil. Trans. Roy. Soc. London, Vol. A247, Pp.
529–551, April 1955. (References)
[2] J. Clerk Maxwell, A Treatise on Electricity and Magnetism, 3rd Ed., Vol. 2.
Oxford: Clarendon, 1892, Pp.68–73.
[3] I. S. Jacobs And C. P. Bean, “Fine Particles, Thin Films and Exchange
Anisotropy,” In Magnetism, Vol. Iii, G. T. Rado And H. Suhl, Eds. New York: Academic,
1963, Pp. 271–350.
[4] K. Elissa, “Title of Paper If Known,” Unpublished.
311
[5] R. Nicole, “Title of Paper with Only First Word Capitalized,” J. Name Stand.
Abbrev., In Press.
[6] Y. Yorozu, M. Hirano, K. Oka, And Y. Tagawa, “Electron Spectroscopy Studies
on Magneto-Optical Media and Plastic Substrate Interface,” Ieee Transl. J. Magn. Japan, Vol.
2, Pp. 740–741, August 1987 [Digests 9th Annual Conf. Magnetics Japan, P. 301, 1982].
[7] M. Young, The Technical Writer’s Handbook. Mill Valley, Ca: University
Science, 1989.
312
Kronik Böbrek Hastalığının Makine Öğrenmesi Teknikleri ile Sınıflandırılması
Mustafa İlker ERDURSUN1, Hasan ERBAY2, Ömer Faruk AKMEŞE1, İbrahim DOĞAN1
1
2
Bilgisayar Teknolojileri, Hitit Üniversitesi, Çorum
Bilgisayar Mühendisliği, Kırıkkale Üniversitesi, Kırıkkale
Abstract: Teknolojinin ilerlemesi ile birlikte birçok veri dijital ortamlarda kayıt altına alınarak
büyük veri yığınları ortaya çıkmıştır. Veri madenciliği sayesinde bu büyük veri yığınlarının
içinden anlamlı ve yararlı bilgilerin ortaya çıkarılması için çalışmalar yapılmaktadır. Özellikle
büyük veri yığınlarını analiz etmede klasik analiz yöntemlerinin yetersiz kalması veri
madenciliği yöntemlerinin önemini arttırmıştır. Her dönemde olduğu gibi günümüzün en
önemli araştırma alanı olan tıp alanında da sürekli olarak hastalara ait veriler artarak kayıt altına
alınmaktadır. Kayıt altına alınan veriler bazen tek başına anlamsız gibi görünürken diğer
verilerle birlikte bütünsel olarak analiz edildiğinde gizli kalmış önemli bilgiler elde
edilebilmektedir. Bu değerli bilgiler, sağlık sektörünün gelişmesine ve doktorların daha doğru
bir şekilde teşhis verebilmesine yardımcı olmaktadır. Bu çalışmada, Kronik Böbrek Hastalığı
(KBH) veri seti üzerinde analiz yapılmıştır. Farklı modeller oluşturularak, bu modellerin veri
üzerindeki tahmin sonuçları karşılaştırılmış ve bu sonuçlara bağlı olarak veriler üzerinde hangi
modelin daha iyi sonuç verdiği belirtilmiştir.
Keywords: Makine Öğrenmesi, Tıbbi Veri Madenciliği, Hastalık Tahmini, Sınıflandırma.
REFERENCES
[1] Meganathan, D., & Marudachalam, N. International Journal of Engineering Sciences
&Management Research.
[2] Raghavendra, S., & Indiramma, M. (2015). Classification and Prediction Model Using
Hybrid Technique For Medical Datasets. Analysis, 127(5).
[3] Singla, M., & Singh, K. Heart Disease Prediction System Using Data Mining Clustering
Techniques.
313
[4] Joshi, S., Shenoy, D., Rrashmi, P. L., Venugopal, K. R., & Patnaik, L. M. (2010, February).
Classification of Alzheimer's Disease And Parkinson's Disease By Using Machine Learning and Neural
Network Methods. Machine Learning And Computing (ICMLC), 2010 Second International Conference
(Pp. 218-222). IEEE.
[5] Peter, T. J., & Somasundaram, K. (2012, March). An Empirical Study on Prediction of Heart
Disease Using Classification Data Mining Techniques. In Advances İn Engineering, Science And
Management (ICAESM), 2012 International Conference (Pp. 514-518). IEEE.
[6] Robu, R., & Hora, C. (2012, June). Medical Data Mining With Extended WEKA. In
Intelligent Engineering Systems (INES), 2012 IEEE 16th International Conference on (Pp. 347-350).
IEEE.
[7] Dewan, A., & Sharma, M. (2015, March). Prediction Of Heart Disease Using A Hybrid
Technique İn Data Mining Classification. In Computing For Sustainable Global Development
(Indıacom), 2015 2nd International Conference on (Pp. 704-706). IEEE.
[8] Soni, J., Ansari, U., Sharma, D., & Soni, S. (2011). Predictive Data Mining For Medical
Diagnosis: An Overview Of Heart Disease Prediction. International Journal Of Computer Applications,
17(8), 43- 48.
[9] Domingos, P., & Pazzani, M. (1997). On The Optimality of The Simple Bayesian Classifier
Under Zero-One Loss. Machine Learning, 29(2-3), 103- 130.
[10] Shrivas, A. K., & Yadu, R. K. (2017). An Effective Prediction Factors For Coronary Heart
Disease Using Data Mining Based Classification Technique. International Journal on Recent And
Innovation Trends İn Computing And Communication, 5(5), 813-816.
[11] Le Cessie, S., & Van Houwelingen, J. C. (1992). Ridge Estimators İn Logistic Regression.
Applied Statistics, 191-201.
[12] Tong, S., & Chang, E. (2001, October). Support Vector Machine Active Learning For
İmage Retrieval. In Proceedings Of The Ninth ACM İnternational Conference on Multimedia (Pp. 107118). ACM.
[13] Hearst, M. A., Dumais, S. T., Osuna, E., Platt, J., & Scholkopf, B. (1998). Support Vector
Machines. IEEE Intelligent Systems And Their Applications, 13(4), 18-28.
[14] Lin, C. F., & Wang, S. D. (2002). Fuzzy Support Vector Machines. IEEE Transactions on
Neural Networks, 13(2), 464-471.
314
[15] Alam, M. S., & Vuong, S. T. (2013, August). Random Forest Classification For Detecting
Android Malware. In Green Computing And Communications (Greencom), 2013 IEEE And Internet Of
Things (İthings/Cpscom), IEEE International Conference on And IEEE Cyber, Physical And Social
Computing (Pp. 663-669). IEEE.
[16] Pal, M. (2005). Random Forest Classifier For Remote Sensing Classification. International
Journal Of Remote Sensing, 26(1), 217-222.
315
Survey on Dynamic Bayesian Network Software Tools
Hüseyin ÇAMBAŞI1, Özgür KURU1, Mehmet Fatih AMASYALI1, Sofiene TAHAR2
1
Department of Computer Engineering, 2Department of Electrical and Computer Engineering
1
Yıldız Technical University, TURKEY, 2Concordia University, CANADA
Abstract: Bayesian networks are probabilistic graphical representations which are used to build
models from data and/or expert opinion. They can be utilized for a wide range of tasks including
prediction, anomaly detection, diagnostics, automated insight, reasoning, decision making, etc.
Dynamic Bayesian Networks (DBN) are extensions of Bayesian networks with temporal
support, which can be used to model systems that dynamically change by the time. Nowadays,
DBNs are utilized in a wide range of applications including robotics, data mining, speech
recognition, digital forensics, protein sequencing, and bioinformatics. Several software tools
exist in the public as well as commercial domains that support modelling and simulation of
DBNs. However, these DBN software tools differ in terms of features support, ease of use,
documentation, user’s community, etc. Therefore, it has become important to establish various
metrics for selecting the proper software tools for creating and simulating DBNs, such as cost,
licensing, GUI, built-in support for inference algorithms, structural learning, data types, etc.
The goal of this survey is the evaluation and comparison of existing software tools for building
DBNs based on a set of users centered criteria.
Keywords: Dynamic Bayesian Networks, Modeling, Simulation, Software, Tools.
REFERENCES
[1] N. Friedman, D. Geiger, And M. Goldszmidt, Bayesian Network Classifiers, Machine
Learning, 29(2-3):131-163, 1997.
[2] V. Mihajlovicand M. Petkovic, Dynamic Bayesian Networks: A State of The Art, Technical
Report, Electrical Engineering, Mathematics and Computer Science, University of Twente, the
Netherlands, 2001.
[3] K. Murphy, Software For Graphical Models: A Review. Isba Bulletin (December 2007).
316
[4] K. Murphy, Software Packages For Graphical Models, Software Packages for Graphical
Models. [Online]. Available: Http://People.Cs.Ubc.Ca/ Murphyk/Software/Bnsoft.Html. [Accessed: 18Apr-2019].
[5] M.A.Mahjoub And K. Kalti, Software Comparison Dealing with Bayesian Networks,
Advances in Neural Networks Lncs 6677 Pp.168-177, 2011.
[6] Bayesialab, Bayesia, Https://Www.Bayesialab.Com/, 2019, Accessed April 28, 2019.
[7] Bayesfusion, Bayesfusion Ltd, Https://Www.Bayesfusion.Com/, 2019, Accessed April 28,
2019.
[8] Bayes Server, Bayes Server Ltd, Https://Www.Bayesserver.Com/, 2019, Accessed April 28,
2019.
[9] Netica, Norsys Software Corp, Https://Www.Norsys.Com/Netica.Html, 2019, Accessed
April 28, 2019.
[10] Hugin Lite, Hugin Expert A/S, Https://Www.Hugin.Com/, 2019, Accessed April 28, 2019
[11] Bnt, Https://Github.Com/Bayesnet/Bnt, 2018, Accessed April 28, 2019.
[12] Banjo, Https://Users.Cs.Duke.Edu/ Amink/Software/Banjo/, 2010, Accessed April 28,
2019.
[13] Amidst, Http://Www.Amidsttoolbox.Com/, 2018, Accessed April 28, 2019.
[14] Bnlearn, Http://Www.Bnlearn.Com/, 2019, Accessed April 28, 2019.
[15] Infer.Net, Https://Dotnet.Github.İo/İnfer/, 2019, Accessed April 28, 2019.
[16] Mocapy++, Https://Sourceforge.Net/Projects/Mocapy/, 2015, Accessed April 28, 2019.
[17] Libpgm, Https://Pythonhosted.Org/Libpgm/, 2012, Accessed April 28, 2019.
[18] Pgmpy, Http://Pgmpy.Org/, 2016, Accessed April 28, 2019.
317
Determination of Factors Affecting Employee Productivity
Taner ERSÖZ1, Filiz ERSÖZ2, Nurdan KALELİ2, Burçin ATICI2
1
Department of Actuary and Risk Management, 2Department of Industrial Engineering
Karabük University, TURKEY
Abstract: Productivity is the relationship between the output produced by a production or
service system and the input used to create this output. High efficiency is to produce more with
the same amount of resources or more output with the same input. However, it is accepted that
efficiency and working life quality are closely linked. The efficiency of the enterprises is
important for the society and the country in which they carry out their activities as well as for
themselves. While the productivity of the enterprises is reflected in the decrease in the costs,
increase in the profit, the efficiency of the enterprises has reflections on the new investments,
increased added value, the welfare of the society and employment in terms of society and
country. It is also important to measure the efficiency that is so important in order to make it
sustainable in enterprises. The productivity measured continuously in enterprises will enable
managers to make correct and rational decisions, to see problems on time and to produce
solutions. The purpose of this study is to determine the factors affecting the productivity of the
employees in the steel industry. The study was examined under five titles. These are:
“Demoraphic Information”, “Economic Factors”, “Physical and Ergonomic Factors”, “PsychoSocial Factors” and “Risk Factors”. According to the findings of the study, the effect of factors
on the employees will be determined and improved findings will be presented to the
management of the enterprise.
REFERENCES
[1] Birdal, İlker & Nilgün Aydemir; Yönetim Teorileri, Sistem Yayıncılık, İstanbul, 1992.
[2] Bingöl, Dursun. Personel Yönetimi ve Beşeri İlişkiler. Erzurum: Atatürk Üniversitesi Basımevi,
1990.
318
[3] Byars, Lloyd L.; Concepts of Strategic Management, 3rd Edition, Harper Collins Publishers, New
York, 1992.
[4] Can, Halil. Organizasyon ve Yönetim, Ankara: Siyasal Kitabevi, 1999.
[5] Düren, Zeynep. 2000’li Yıllarda Yönetim, İstanbul: Alfa Basım Yayın Dağıtım A.Ş, 2000
[6] Eren, Erol Yönetim ve Organizasyon, İstanbul: Beta Basım Yayım Dağıtım A.Ş, 1998. İşgören
Verimliliğini Etkileyen Faktörlerin İncelenmesine 251 C.13, S.3
[7] Hersey Paul & Kenneth H. BLANCHARD, Management and Organizational Behaivor,
Englewood Cliffs: Prentice Hall, 1979.
[8] Luthans, Fred. Organizational Behaivor. U.S.A.: Mc Grow Hill, 1995.
[9] Özdamar, Serpil. “İnsan Gücü Potansiyelimizin En verimli Biçimde Değerlendirilmesi Öncelikli
Hedeflerimiz Arasındadır”, İşveren Dergisi. 8, 1998.
[10] Özçer, Sema. Verimliliğe Etkileri Açısından Sanayi İşletmelerinde Örgüt Yapıları ve Liderlik
Biçimleri, Ankara: MPM Yayınları, 1995.
[11] Preffer, Jeffery Rekabette üstünlüğün Sırrı: İnsan.Çeviren: Sinem Gül, İstanbul: Sabah
Kitapçılık San. Ve Tic. A.Ş, 1995.
[12] Sabuncuoğlu, Zeyyat & Melek Tüz, Örgütsel Psikoloji, Bursa: Alfa Basım Yayım, 1998
[14] Sapancalı, Faruk. Çalışanların Güdülenmesinde Kullanılan Özendirici Araçlar, Verimlilik
Dergisi, 1993.
[15] Sherman, A. W. Herbert J. Churden & J.R., Personnel Management, Ohio: South Western
Publishing Co.
[16] Storey John & Sisson Keith, Managing Human Resources and Industrial Relations, Buckingam:
Open University Pres, 1995.
[17] Uğur, Adem “Türkiye’de İşgücü Verimliliğini Etkileyen Sosyo-Kültürel Faktörlerin Önemi”, I.
Verimlilik Kongresi, 1991.
[18] Özkılıç, Ö. (2014), Risk Değerlendirmesi (Atex Direktifl eri-Patlayıcı OrtamlarBüyük
Endüstriyel Kazaların Önlenmesi ve Etkilerinin Azaltılması-Kantitatif Risk Değerlendirme). Ankara:
Türkiye İşveren Sendikaları Konfederasyonu Yayınları.
[19] Ocaktan, E. (2014). Meslek Hastalıkları. Çalışan Sağlığı ve Güvenliği Eğitimi ve Çalıştayı. 0205 Aralık. Ankara.
319
Adem Aksan