| Summary: | 碩士 === 國立雲林科技大學 === 資訊管理系碩士班 === 99 === Fuzzy time series have in recent years drawn many scholars'' attention due to their ability can handle the time series data with incomplete, imprecise and ambiguous pattern. However, most traditional time series models employed only single variable (stock index) in forecasting, yet ignored some factors that would also affect the stock volatility. Therefore, this paper proposes a novel forecasting model using multi-factor fuzzy time series model to forecast stock index. Multi-factor fuzzy time series model is composed of three main components: stock index, trading volume and interactions between two stock markets. In order to evaluate the performance of the proposed model, the transaction records of NASDAQ (National Association of Securities Dealers Automated Quotations)、TAIEX (Taiwan Stock Exchange Capitalization Weighted Stock index)、HSI (Hang Seng Index) from 1997 to 2004, and the root mean square error (RMSE) are used as experimental dataset and evaluation criterion, respectively. Then, we adopt three recent fuzzy time-series models: Chen’s (1996), Yu’s (2005), Teoh’s (2008), Support vector regression (SVR) and General Regression Neural Network (GRNN) as compared methods. The results show that the proposed model outperforms the listed models in accuracy for forecasting stock index.
|
|---|
