Forecasting KOSPI by weighted average defuzzification based on NEWFM

Fuzzy neural networks have been successfully applied to generate predictive rules for stocks forecasting. This paper presents a methodology for forecasting the daily Korea composite stock price index (KOSPI) based on the neural network with weighted fuzzy membership functions (NEWFM) and time series of KOSPI based on the defuzzyfication of weighted average method (The fuzzy model suggested by Takagi and Sugeno in 1985). NEWFM is a new model of neural networks to improve forecasting accuracy by using self adaptive weighted fuzzy membership functions. The degree of classification intensity is obtained by bounded sum of weighted fuzzy membership functions extracted by NEWFM, and then weighted average defuzzification is used for forecasting KOSPI. In this paper, the Haar wavelet function is used as a mother wavelet. A set of five extracted coefficient features of the Haar WT are presented to forecast KOSPI. The total number of samples is 2928 trading days, from January 1989 to December 1998. About 80% of the data is used for training and 20% for testing. The result of classification rate is 58.0034%. The implementation of the NEWFM demonstrates an excellent capability in the field of stocks forecasting.