Consistency of Clustering Between Optimal Fuzzy Equivalence Relation and C-Means
Zhonglin Chai · 2024
In cluster analysis, there are two important kinds of fuzzy clustering methods. One is fuzzy C-means clustering, in which the model proposed by Bezdek is a typical representative. Another kind is the clustering based on similarity relations, such as transitive closure method. What is the inherent connection between the two kinds of clustering methods? This paper studies this problem based on their two representative models. Through theoretical analyses and a case test it finds that they are consistent in cluster analysis. This not only enriches clustering theory, but also has practical value. In addition, this paper also makes some discussions on the clustering based on similarity relations, in order to make the methods of this kind better applied to cluster analysis.