The most challenging issues in cluster analysis are to ratify clustering results to produce optimal number of cluster for a dataset. Customary validity indices are geometry based, and these indices face a vital problem where the resultant index value either increases or decreases as cluster count inflates. This paper exhorts a cluster validity index for rough fuzzy c-means clustering algorithm called rough fuzzy Bayesian like validation method which ischolar_mains on probabilistic metric. Maximum Bayesian score stipulates optimal number of cluster. The proposed measure is been experimented for synthetic and diverse UCI datasets. This recommended scheme brings out optimal number of cluster for enormous UCI datasets than the prevailing customary validity indices.
Keywords
Bayesian Like Validation Method, Cluster Analysis, Probability, Rough Fuzzy C-Means, UCI Datasets.
User
Information