Forecasting Stock Price Index Using Fuzzy Time-Series Based on Rough Set

Abstract

Fuzzy time-series have been utilized to make predictions in various areas such as stock price forecasting, academic enrollments and weather. In the forecasting processes, Fuzzy Logical Relation (FLR) is the one of critical factors to influence forecasting accuracy. Therefore, in this paper, we propose a new fuzzy time-series method, which employs rough set theory to mine FLR in time-series and the adaptive expectations model to tune forecasting results. In the empirical analysis, we use a ten-year period of TAIEX (Taiwan Stock Exchange Capitalization Weighted Stock Index) closing prices as experimental datasets and two fuzzy time-series methods, Chen's (1996) and Yu's (2004) methods, as comparisons models. The experimental results shows that propose method outperforms the listing methods.