Research And Implementation Of Fuzzy Clustering Algorithm

This paper will illustrate "clustering analysis" thoroughly. Cluster is a process that assorts things by their similarity. There is no adviser in this process, so it is a non-supervised classification. "Clustering analysis" research and process assort things by mathematical means. Traditional Clustering analysis assorts things strictly; therefore the limit of the classification is very clearly. But in fact most of the things have no obvious attribute by each; their limit is vague, as a result soft classification is a better way to process them. Professor Zadeh introduced the theory of fuzzy sets, which offer a powerful means to solve the problem. People begin to use fuzzy way to deal with clustering problem, and call it "fuzzy clustering analysis" ."Fuzzy clustering analysis" contains three steps. The first is data standardization; the second is to establish fuzzy similar matrix; the third is clustering. This paper will research two arithmetic of the Fuzzy clustering analysis by test. And then we will analyze the "transitive closure" arithmetic through two procedures: first, by comprehending characteristic of the transitive closure arithmetic to analyze its defect; second, using the concept of "optimal fuzzy equivalent matrix" . And then we will give a new Fuzzy clustering analysis arithmetic. The name is "approximate optimal equivalent matrix" arithmetic.Finally, the paper will accomplish Fuzzy clustering analysis program by VB. It is significant to use data to validate it.