VALIDITY STUDIES IN CLUSTERING METHODOLOGIES

Clustering algorithms tend to generate clusters even when applied to random data. This paper provides a semi-tutorial review of the state-of-the-art in cluster validity, or the verification of results from clustering algorithms. The paper covers ways of measuring clustering tendency, the fit of hierarchical and partitional structures and indices of compactness and isolation for individual clusters. Included are structural criteria for validating clusters and the factors involved in choosing criteria, according to which the literature of cluster validity is classified. An application to speaker identification demonstrates several indices. The development of new clustering techniques and the wide availability of clustering programs necessitates vigorous research in cluster validity. © 1979.