Based On Hierarchical Fuzzy Clustering Algorithm

Clustering is one of the important tasks in the field of data mining. Fuzzy clustering analysis which introduces the theory of fuzzy sets, provides the capability that can be used to deal with real fuzzy datasets. And it has been widely used in many fields.Based on the analysis of the classic fuzzy c-means clustering algorithm and its variation, which belong to the fuzzy clustering algorithms based on partitions, the paper proposes two fuzzy clustering algorithms and a module description which are based on hierarchies. Among of them the secondly fuzzy clustering algorithm based on the centers of dense clusters which introduces some quality measurement about clustering, avoids the number of clusters to be the input parameter. To improve the capability of algorithms dealing with fuzzy datasets, propose the hierarchical fuzzy clustering algorithm based on dynamic modeling. On the base of FCM and other two algorithms mentioned above, the paper describes the procedure of fuzzy clustering on datasets based on the hierarchies in modules.SFCC uses the most possible cluster number as the input parameter other than the direct cluster number because of the quality measurement and does not prefer to find the global and similar size clusters. Based on the k-nearest neighbor graph, DMFC analyses and clusters the original partition.Implementing FCM, SFCC and DMFC these three algorithms and respectively verifying them on various datasets, the clustering results and the analysis of time complication indicate that the algorithms this paper proposes are valid in dealing with fuzzy datasets.