FORECASTING FINANCIAL DATA: A COMBINED MODEL OF FUZZY NEURAL NETWORK AND STATISTICS

摘要

In this paper, we apply the ARMA/ARCH methodology to develop forecasting models and compare their forecast accuracy with a class of novel hybrid fuzzy logic RBF neural network models. The used novel approach deals with nonlinear estimate of various RBF NN-based ARMA/GARCH methodologies. Our results show that the proposed approach achieves better forecast accuracy on the validation dataset than most available techniques.