点击次数：

论文类型：期刊论文

发表时间：2013-01-01

发表刊物：ICIC Express Letters

收录刊物：EI、Scopus

卷号：7

期号：9

页面范围：2601-2607

ISSN号：1881803X

摘要：Partitioning the universe of discourse and determining intervals are very important issues for forecasting in fuzzy time series. Equal length intervals used in most existing literatures are convenient but subjective to partition the universe. In this paper, the purpose of our study is to determine effective intervals with unequal length to improve forecast accuracy. A new method to determine intervals with unequal length is proposed and these intervals carry well-defined interpretability. Firstly, using fuzzy c-means clustering algorithm we calculate the prototypes of data set, and secondly, we obtain some subsets according to the distribution of the prototypes. At last, we get the effective intervals by information granules. The proposed method is evaluated using experimental data and its performance is contrasted with the equal length intervals available in the literature. Empirical results show that the proposed method can greatly improve forecast a ccuracy. ? 2013 ICIC International.