SOME NEW INDEXES OF CLUSTER VALIDITY

We review two clustering algorithms (hard c-means and single linkage) and three indexes of crisp cluster validity (Hubert's statistics, the Davies-Bouldin index, and Dunn's index). We illustrate two deficiencie s of Dunn's index which make it overly sensitive to noisy clusters and propose several generalizations of it that are not as brittle to outl iers in the clusters. Our numerical examples show that the standard me asure of interset distance (the minimum distance between points in a p air of sets) is the worst (least reliable) measure upon which to base cluster validation indexes when the clusters are expected to form volu metric clouds. Experimental results also suggest that intercluster sep aration plays a more important role in cluster validation than cluster diameter. Our simulations show that while Dunn's original index has o perational flaws, the concept it embodies provides a rich paradigm for validation of partitions that have cloud-like clusters. Five of our g eneralized Dunn's indexes provide the best validation results for the simulations presented.