Group for Research in Decision Analysis


Maximum Split Single Cluster Clustering


Consider a set O of N entities and a matrix D = (dk ) of dissimilarities between pairs of entities {Ok, O} of O. The split of a subset (or cluster) C of O is the minimum dissimilarity between an entity of C and one of O C. A (N2) algorithm is proposed for finding a maximum split cluster C of m entities for all m = 1,2,...,N - 1.

, 10 pages