A cluster validity index for hard clustering

Abstract

This paper describes a new cluster validity index for the well-separable clusters in data sets. The validity indices are necessary for many clustering algorithms to assign the naturally existing clusters correctly. In the presented method, to determine the optimal number of clusters in data sets, the new cluster validity index has been used. It has been applied to the complete link hierarchical clustering algorithm. The basis to define the new cluster validity index is founding of the large increments of intercluster and intracluster distances, when the clustering algorithm is performed. The maximum value of the index determines the optimal number of clusters in the given set simultaneously. Obtained results confirm very good performances of the proposed approach. © 2012 Springer-Verlag Berlin Heidelberg.