Clustering Tendency Problem in Pattern Analysis

Abstract

Determining the structure of multi-dimensional patterns is an important problem in exploratory pattern analysis. Clustering methods have been used extensively for this purpose. However, clustering algorithms will locate and specify clusters in data even non are present. It is therefore appropriate to measure the clustering tendency or randomness of a pattern set before subjecting it to a clusterinb algorithms. We survey the work that has been done in developing measures of clustering tendency. Methods based on quadrats, spatial model fitting, distribution of interpoint distances are reviewed with special attention to distance-based methods that use sampling origins. One of the goals of this study is to assess their applicability to high dimensional patterns. The successes and failures of these methods are discussed as well as suggestions and directions for future study.