KESKİN BENLİ, Yasemin Gazi Üniversitesi - Endüstriyel Sanatlar Eğitim Fakültesi - İşletme Eğitimi Bölümü, Turkey , GÜNERİ TOSUNOĞLU, Nuray Gazi Üniversitesi - Ticaret ve Turizm Eğitim Fakültesi - Bilgisayar Uygulamaları Eğitimi Bölümü, Turkey

EVALUATE THE MORGAN STANLEY CAPITAL INTERNATIONAL INDEX OF THE EUROPEAN UNION COUNTRIES AND FORECAST BY ARTIFICAL NEURAL NETWORKS

36331

Financial forecasting is an important issue that providing an idea about the future of the economy according to both academics and investors. In this study, we aimed to evaluate and forecast the Morgan Stanley Capital International index of the fourteen European Union countries. Artificial neural networks is chosen for forecasting. Data period is 31 December 1987-31 October 2013. The artificial neural network architecture is determined separately for each index. 12 lagged time series is used in the study because of the data is monthly. The number of neurons in the input layer is 12. For the selection of the number of neurons in the hidden layers, number of neurons that vary between 1 and 12 of the 12 cases are evaluated for each country. The model having a minimum RMSE and MAE value was created network architecture. For each country, the best weight values were calculated. As performance criteria the RMSE and MAE was selected.

72

Stock Exchange , European Union Countries , Time Series Analysis , Artificial Neural Networks , Forecasting.

Journal Of Economics an‎d Administrative Sciences

87