WEIGHTING AND SELECTION OF VARIABLES FOR CLUSTER-ANALYSIS

One of the thorniest aspects of cluster analysis continues to be the w eighting and selection of variables. This paper reports on the perform ance of nine methods on eight ''leading case'' simulated and real sets of data. The results demonstrate shortcomings of weighting based on t he standard deviation or range as well as other more complex schemes i n the literature. Weighting schemes based upon carefully chosen estima tes of within-cluster and between-cluster variability are generally mo re effective. These estimates do not require knowledge of the cluster structure. Additional research is essential: worry-free approaches do not yet exist.