A novel cluster validity index for fuzzy clustering based on bipartite modularity

A novel cluster validity index whose implementation is based on the membership degrees and improved bipartite modularity of bipartite network is proposed for the validation of partitions produced by the fuzzy c-means (FCM) algorithm. FCM algorithm is employed to group the dataset in order to obtain the membership degree of samples. Then, a weighted bipartite network is constructed by samples and centroids of each cluster. This allows the introduction of a new measurement for optimizing the numbers of clusters for fuzzy partitions. The proposed index utilizes the optimum membership as its global property and the modularity of bipartite network as its local independent property. The proposed index is compared with a number of popular validation indices on fifteen datasets. The experimental results show that the effectiveness and reliability of the proposal is superior to other indices.