Analyzing Performance of Artificial Neural Networks by Taguchi Methods: Forecasting Stock Market Prices

Öz

A wide variety of statistically based forecasting methods have been developed depending on the nature of the problem concerned. An Artificial Neural Network (ANN), one of the artificial intelligence techniques, is considered to be a powerful tool for recognition, classification, forecasting, and optimization. The design of an ANN architecture has been widely recognized as an important issue in the literature. In this paper, Taguchi method was used to analyze tha factors that affect the performance of an ANN and to determine appropriate values of the factors regarding to improving the performance of an ANN. A stock, namely Koç Holding, has been chosen from IMKB for application purpose and the optimized ANN has bee applied to forecast prices of the selected stock using data in the period of Jan96-Dec 01. In order to determine efficiency of the approach and to emphasize the necessity of performance optimization in ANN, the result obtained was compared with a randomly designed ANN and the multiple regression model.

Anahtar Kelimeler

• Taguchi Method
• Artificial Neural Network
• Regression Analysis
• Forecasting
• Stock

Yapay Sinir Ağı Performansına Etki Eden Faktörlerin Analizinde Taguchi Yöntemi: Hisse Senedi Fiyat Tahmini Uygulaması

Öz

İlgilenilen problemin yapısına bağlı olarak istatistiksel pek çok tahminleme yöntemi geliştirilmiştir. Bir yapay zeka tekniği olan Yapay Sinir Ağı (YSA); tanıma, sınıflandırma, tahminleme ve eniyileme konularında kullanılan etkili bir tekniktir. YSA’da model belirleme problemi literatürde önemli bir konu olarak ele alınmaktadır. Bu çalışmada, YSA performansını etkileyen faktörlerin analizi ve performansını iyileştiren uygun değerlerin belirlenmesinde taguchi yöntemi kullanılmıştır. Uygun faktör değerleri belirlenmiş YSA, İstanbul Menkul Kıymetler Borsası’nda (İMKB) işlem gören Koç Holding hisse senedinin fiyat tahmini problemine, Ocak 96- Aralık 01 dönemi için derlenen veriler dikkate alınarak uygulanmıştır. Yaklaşımın etkinliğini belirlemek ve YSA’da model seçiminin önemini vurgulamak amacıyla, elde edilen sonuç rassal olarak tasarlanmış bir YSA ve çoklu doğrusal regresyon modeli ile karşılaştırılmıştır.

Anahtar Kelimeler

• Taguchi Yöntemi
• Yapay Sinir Ağı
• Regresyon Analizi
• Tahminleme
• Hisse Senedi

Kaynakça

1. AIKEN, M. (1999), Using a Neural Network to Forecast Inflation, Industrial Management and Data Systems, 99:7, 296-301.
2. ALTUĞ, S. (1994), Price Prediction in İMKB Using Neural Networks, MBA Thesis, Bilkent University, Ankara.
3. ANAGÜN, A.S. (1999), Bilgi Güvenliğinin Sağlanmasında Kullanıcı Özelliklerine Dayalı Bir Yapay Sinirsel Ağ Yaklaşımı, Endüstri Mühendisliği, 10: 4, 3-11.
4. ANAGÜN, A.S., LIUO, Y.H.A. (1993), A Neural Network Application for Apnea Recognition: A Preliminary Study, ASME Intelligent Engng System Through Artificial Neural Networks, 3, 321-326.
5. ANTONY J., ROY, R.K. (1999), Improving the Process Quality Using Statistical Design of Experiments: A Case Study, Quality Assurance, 6, 87-95.
6. BAILEY, D., THOMPSON D. (1990), How to Develop Neural Network Applications, Al Expert, 38-47.
7. BALABAN, E., CANDEMİR, H.B., KUNTER, K. (1996), İstanbul Menkul Kıymetler Borsasında Aylık Dalgalanma Tahmini, Sermaye Piyasası ve İMKB Üzerine Çalışmalar, İşletme ve Finans Yayınları, Ankara.
8. BERBEROĞLU, N., ARSLAN, S., AFŞAR, M. (1992), Hisse Senetlerinde Değerleme Yöntemleri ve Türkiye’de Hisse Senetlerinin Fiyatlarını Belirleyen Faktörlerin Analizi, Anadolu Üniversitesi İktisadi ve İdari Bilimler Fakültesi Dergisi, 10, 1-34.

9. BOSE, N.K., LIANG, P. (1996), Neural Network Fundamentals with Graphs, Algorithms, and Applications, Mcgraw-Hill, New York.
10. BURR, D. J. (1988), Experiments on Neural Net Recognition of Spoken and Written Text, IEEE Transactions on Accoustics, Speech, and Signal Processing, 36:7, 1162-1168.
11. CAUDILL, M. (1993), What is a Neural Network?, Al Expert, 7-14.
12. CAUDILL, M. (1988), Neural Network Premier, Part-I Al Expert, 2-7.
13. DANIELS, H., KAMP, B. (1999), Application of MLP Networks to Bond Rating and House Pricing, Neural Computation and Applications, 8, 226-234.
14. ENKE., D., DIWE, P., VAITIANATHASAMY, S. (2000), Factorial Design for Developing Feed Forward Neural Network Architectures, ASME Intelligent Engng System Through Artificial Neural Networks, 10, 109-114.
15. FARAWAY, J.,CHATFIELD. C. (1998), Time Series Forecasting with Neural Networks: A Comparatve Study Using the Airline Data, Application Statistics, 47: 2, 231-250.
16. FOWLKES, W.Y. CREVELİNG, C.M. (1995), Engineering Methods for Robust Product Design Using Taguchi Methods in Technology and Product Development, Canada: Addison-Wesley Publishing Company.
17. FU, L.M. (1994), Neural Networks in Computer Intelligence, New York: Mcgraw-Hill.
18. GREFENSTETTE, J. (1986), Optimization of Control Parameters for Genetic Algorithms, IEEE Transactions on Systems, Man, and, Cybemetics, 16, 122-128.
19. HARRINGTON, D. (1987), Modern Portfolio Theory, New Jersey: Prentice Hall.
20. KLIMASAUSKAS, C. C. (1989), An Introduction to Neural Networks, Part III: Training a Neural Network, PC AI, 20-24.
21. LAW, R. (1998), Room Occupancy Rate Forecasting: a Noural Network Approach, International Journal of Contemporary Hospitality Management, 10: 6, 234-239.
22. MANSON, E., WANG, Y.J. (1990), Introduction to Computing and Learning in Artificial Neural Networks, European Journal of Operational Research, 47:1, 1-28.
23. MCLAUCHLAN, R.A., WECKMAN, G.R., PALLERLA, S., VEGALETI, V. (1999), Predicting Student Academic Success in the Engineerimg Curriculum at Texas A&M University Kingville Using Neural Networks, ASME Intelligent Engng System Through Artificial Neural Networks, 9, 1183-1188.
24. NASIR, M.L., JOHN, R.I., BENNETT, S.C., RUSSELL, D.M. (2001), Selecting the Neural Network Topology for Student Modelling of Prodiction of Corporate Bankruptcy, Campus Wide Information Systems, 18: 1, 13-22.
25. OLIVERIA, K.A., VANNUCCI, A, SILVA, E.C. (2000), Using Artificial Neural Networks to Forecast Chaotic Time Series, Physica, a, 284, 393-404.
26. ÖZÇAM, F. (1996), Teknik Analiz ve İstanbul Menkul Kıymetler Borsası, Sermaye Piyasası Kurulu, Ankara.
27. RAGGAD, B.G. (1996), Neural Network Technology for Knowledge Resource Management, Management Decision, 34: 2, 20-24.
28. ROSS, P.J. (1988), Taguchi Techniques for Quality Engineering, New York: McGraw-Hill.
29. SARI, Y. (1992), Borsada Teknik Analiz, İstanbul: Scala Yayıncılık&Tanıtım A.Ş.
30. SELER, İ.T. (1996), Haftanın Günleri: İMKB’ye Etkileri Üzerine Bir İnceleme, Sermaye Piyasası ve İMKB Üzerine Çalışmalar, İşletme ve Finans Yayınları, Ankara.
31. SIMPSON, P.K. (1990), Artificial Neural Systems: Foundations, Paradigms, Applications, and Implementations, New York: Pergamon Press.
32. SİYAHİ, B., ANAGÜN, A.S. (1998), Sismik Sıvılaşma Potansiyelinin Yapay Sinir Ağları ile Belirlenmesi Zemin Mekaniği ve Temel Mühendisliği Yedinci Ulusal Kongresi Bildiriler Kitabı, 552-561.
33. VENUGOPAL, V., BAETS, W. (1994), Neural Networks and Statistical Techniques in Marketing Research: A Conceptual Comparison, Marketing Intelligence and Planning, 12: 7, 30-38.
34. ZARUDA, J. M. (1992), Introduction to Artificial Neural Systems, St. Paul: West Publishing.