期刊名称：International Journal of Computer Technology and Applications
电子版ISSN：2229-6093
出版年度：2012
卷号：3
期号：6
页码：2033-2037
出版社：Technopark Publications
摘要：The Time-Series models have been used to make predictions in whether forecasting, academic enrollments, etc. Song & Chissom introduced the concept of fuzzy time series in 1993[11]. Over the past 19 years, many fuzzy time series methods have been proposed for forecasting enrollments. These methods mainly focus on 3 factors, namely, the universe of discourse, partition of discourse and the defuzzification method. These methods have either used enrollment numbers or difference of enrollments or percentage change as the universe of discourse. And either used frequency density based portioning or ratio-based portioning as the partition of discourse. For defuzzification process, these methods either used Chen’s method or Jilani’s method. The main issue in forecasting is in improving accuracy. But the forecasting accuracy rate of the existing methods is not good enough. This paper proposed a method based on fuzzy time series, which gives the higher forecasting accuracy rate than the existing methods. The proposed method used the percentage change as the universe of discourse; mean based partitioning as the partition of discourse and centroid method for defuzzification. To illustrate the forecasting process, the historical enrollment of University of Alabama is used.
关键词：fuzzy set; fuzzy time series; forecasting; forecast error; time variant model; first order model
