Fuzzy Weighted-Transitional Matrix for Forecasting Time Series Data

This paper proposes a new concept called weighted-transitional matrix which employs two new methods for forecasting time series data. One is expectation method and the other is grade-selection method. This paper uses this concept for forecasting the enrollments of university, which provides important information for administrator in the arrangement of human resource and budget allocated. The average forecasting results of proposed methods are better than existing methods, and perform very stable in the dataset of large variation, and can overcome the drawback that traditional fuzzy time series models can not sufficiently extract hidden information to build forecasting rules.