Research On Attribute Reduction And Fuzzy Clustering With Rough Set Theory

Rough set theory is an important data analysis method which can be used to deal with uncertainties.It can acquire knowledge according to the indistinguishable relationship between data objects without any prior knowledge.Rough set theory provides a new theoretical framework for solving soft computing problems and has a wide application space in data mining.Based on the detailed description of rough set theory,this thesis deeply discusses the application of rough set theory in attribute reduction and fuzzy clustering.In the field of data mining,attribute reduction and clustering have always been the focus of many experts and scholars.With the development of computer technology and communication technology,data sets become more and more complex,and there are a lot of redundant attributes and noise data in the data sets.Attribute reduction can effectively reduce the dimension of data sets,reduce attribute redundancy.Fuzzy clustering is a widely used soft clustering method.The lack of effective processing methods in dealing with noise data often leads to unsatisfactory clustering results.The research on attribute reduction and fuzzy clustering with rough set theory can promote the integration of rough set theory and data mining and improve the effect of data mining.The main research work of this thesis is as follows:1)In the extended model of rough set theory,this thesis introduces an attribute reduction algorithm based on attribute significance,and points out that the algorithm neglects the correlation between conditional attributes in reduction,which often leads to unsatisfactory reduction results.Therefore,an attribute reduction algorithm based on maximum importance and minimum relevance of attributes is proposed.When selecting attributes,the algorithm fully considers the importance of attributes and the relevance between attributes,which can effectively reduce the number of attributes and improve the quality of reduction.2)In order to deal with the problem that the fuzzy C-means clustering algorithm is sensitive to noise points and its convergence speed is too slow,a restraining fuzzy clustering algorithm based on rough set is proposed.According to the related concepts of rough set theory,the membership formula of fuzzy C-means algorithm is redefined to improve the clustering effect.Besides,a restraining factor is set to improve the convergence speed of the algorithm on the premise of guaranteeing the clustering effect.