Font Size: a A A

Fuzzy C-means Cluster Study Of Incomplete Data

Posted on:2018-04-17Degree:MasterType:Thesis
Country:ChinaCandidate:J J ZhengFull Text:PDF
GTID:2348330533966152Subject:Mathematics
Abstract/Summary:PDF Full Text Request
Fuzzy clustering algorithm (FCM) is one of the important algrithms of clustering,the clustering Process id simple and easy to implemet therefore FCM algorithm has been widely used.FCM algorithm is usually only used in clustering analysis of data sets,but incomplete data sets of clusterintg problem is out of putf.In many practical problems,becayse various reasuns lead to the lack of datas,the datas transform into incomplete data sets,therefore,the incomplete data sets of clustering problem is of hot topic.This paper Mainly studies incomplete data set of fuzzy clustering algorithm,for the optimal completion strategy (OCSFCM) and nearest prototype Strategy (NPSFCM)was improved, the deficienricy of the cnmbined TLBO algorihm to choose the best initial center,the main content of this paper includes several aspects:(1)As the optimal completion strategy (OCSFCM) algorithm is sensitive to the choice of initial class center, this article will "teaching and learning" optimization algorithm (TLBO) in combination with DCSFCVM algorithm by using TLBO algorithm for incomplete data set to optimize the initial class center.established TLBO-OCSFCM clustering algorithm.(2)Due to using nearest neighbor algorithm (NPSFCM) for clustering ,the selection of the initial claszs center has great influnence on clustring effect,this articlewill "teaching and learning " combined NPSFCM algorithm.first initial class center ofincomplete data sets is optimized,and then fuzzy clustering was carried out on the incomplete data set,thus established the TLBO-NPSFCM clustering algorithm.(3)The paper will improved TLBO-OCSFCM algorthm and TLBO-NPSFCMalgorithm is applied to UCl data sets of Iris and Wine in the database the empirical analsis,the analysis resulys show that TLBO-OCSFCM and TLBO-OCSFCM algorithm clustering perfmance better than OCSFCM algorithm and NPSFCM clustering effect has obvious improvement of the OCSFCM algorithm.At the same time.the improved TLBO-OCSECM algorithm and TLBO-NPCM algorithm compares clustering performance again .it found that TLBO-OCSFCM algorithm peformance is beter than TLBO-NPSCM algorithm.
Keywords/Search Tags:Partial distance of the Euclusion norm, TLBO, OCSFCM, NPSFCM, TLBO-OCSFCM
PDF Full Text Request
Related items