Determining the Number of Clusters in Any Arbitrary Dataset Containing Reasonably Separated Data Clusters
S. Easwaran · ASME Press eBooks · 2006
One of the essential aspects of cluster analysis is the accurate determination of the number of clusters in any arbitrary dataset with no a-priori information about their grouping structure. A simple yet powerful algorithm is proposed and investigated in this paper to accurately determine the number of clusters in any arbitrary dataset composed of "reasonably separated" data clusters. Verification outcomes are provided to demonstrate satisfactory performance of the algorithm on a wide and representative variety of artificial datasets.