The Multi-scale Algorithm On Fuzzy Time Series

In recent years, in the study of fuzzy time series model, the theory of domain division of good or bad directly affects the accuracy of the prediction accuracy, Therefore, how to reasonable and effective to divide domain has become problem widely concerned by experts and scholars, this paper study of this issue.Chapter1is the introduction part. Mainly introduced the development course of fuzzy time series, and involved the related idea and the main part of this thesis.Chapter2introduces related knowledge. First introduce the basic knowledge of fuzzy set; Followed by the time series of relevant knowledge; Then introduces the meaning of fuzzy time series model and modeling steps of frame structure and the main related; Finally introduced the fuzzy C-average clustering algorithm (FCM).Chapter3mainly discusses the fuzzy time series theory domain division method. Introduced over the years concerning domain division method of fuzzy time series, focuses on several different granularity theory domain methods, such as, through interval and unequal interval differentiate, explain domain division theory’s influence on the prediction accuracy.Chapter4is the main part of this study, discusses the theory of domain division method based on multi-scale ratio. Discuss the theory of domain boundary point selection and optimization of parameters of the method. Comparing the RMSE and MSE values, through the analysis of the instance shows that this improved method is better than the past, the method of that method is effective.