Abstract

Introduction Clustering is one of the most basic and useful methods of data analysis. It involves treating groups of objects as if they were the same, and describing how the groups relate to one another. Clustering summarizes and organizes data, provides a framework for understanding and interpreting the relationships between objects, and proposes a simple description of these relationships that has the potential to generalize to new or di#erent situations. For these reasons, many di#erent clustering models have been developed and used in fields ranging from computer science and statistics to marketing and psychology [see Arabie, Hubert, and De Soete 1996; Everitt 1993; Gordon 1999 for overviews]. Di#erent clustering models can be characterized in terms of the di#erent assumptions they make about the representational structure used to define clusters, and the similarity measures that describe the relationships between objects and clusters. 2.1.1 Representational Assumptions Represe

Citations