A new algorithm for detecting the optimal number of substructures in the data

A new clustering algorithm is proposed. This algorithm uses a weighted Mahalanobis distance (WMD) as a distance metric to perform partitional clustering. This WMD prevents the generation of unusually large or unusually small clusters. Properties of the new algorithm are presented by examining the clustering quality for codebooks designed with the proposed method and two common methods that use Euclidean distance. The new algorithm provides better results than the competing methods on a variety of data sets. Application of this algorithm to the problem of estimation the optimal number of subgroups present in the data set is discussed