A New Cluster Validity Index Using Novel Point-based Compactness Measure

Cluster validation is an important part of clustering process. This is one of the most widely studied problem and a number of methods and indices have been proposed from time to time. The evaluation of clustering results is very important for determining the optimal clustering solution for a given dataset. The most commonly used approaches for cluster validation are based on internal validity indices. In this paper, we propose a new cluster validity index (ARPoints index) for the purpose of cluster validation. The proposed index measures compactness of clusters by using a new ratio of actual and proportionate number of points present in a given space defined in this paper. We conduct a thorough comparison of these indices with the proposed index on a number of datasets which includes shaped and Gaussian-like datasets. Experimental results show that the proposed index performs better than the commonly known indices.