APPLICATIONS OF TAIEX AND ENROLLMENT FORECASTING USING AN EFFICIENT IMPROVED FUZZY TIME SERIES MODEL

摘要

In this study, an improved fuzzy time series model is proposed for solving forecasting problems. The proposed forecasting model is a hybrid of the traditional fuzzy time series and the ratio value. First, the linguistic variable analysis of fuzzy theory in the traditional fuzzy time series is used to observe the uncertain data which can be modeled as fuzzified variables. According to the ratio value in fuzzy logical relationship, the subsequent predicted data will be determined to increase or decrease. Finally, a fuzzy rule table is established for performing prediction. To verify the efficacy of the proposed forecasting model, the university enrollment and the Taiwan Capitalization Weighted Stock Index (TAIEX) forecasting problems are used as experiments. Experimental results show that the proposed model can achieve a better forecasting accuracy than other models.