A multivariate statistical technique for solving classification problems. The object is to sort items into groups such that the degree of association is strong between members of the same cluster and weak between members of different clusters.