An empirical study of fuzzy approach with artificial neural network models


International Journal of Mathematical Models and Methods in Applied Sciences, vol.6, no.1, pp.114-121, 2012 (Scopus) identifier


Time series forecasting based on fuzzy approach by using artificial neural networks is a significant topic in many scientific areas nowadays. Artificial neural network models are sufficient due to their abilities to solve nonlinear problems especially financial researches in recent years. For these reasons, in this paper we made a forecasting study for weekly closed prices of the exchange rate of Turkish Liras (TL) to Euro between 2005 and 2009 which has important effect in economical and industrial areas. We applied the best four networks which are called multilayer perceptron (MLP), radial basis function (RBF) neural network and generalized regression neural network (GRNN) to improve forecasting fuzzy time series with different degrees of membership by using MSE performance measure. Empirical results show that the MLP outperforms others to forecast neural network based-fuzzy time series.