An Empirical Comparison of Methods for Clustering Problems: Are There Benefits from Having a Statistical Model?

This study compares the effectiveness of statistical model-based (MB) clustering methods with that of more commonly used non model-based (NMB) procedures in three important contexts: the traditional cluster analysis problem in which a set of consumer characteristic variables is used to form segments; clusterwise regression, in which response parameters from a regression form the basis of segments, and bicriterion clustering problems, which arise when managers wish to form market segments jointly on the basis of a set of characteristics and response parameters from a regression. If the manager’s primary objective is to forecast responses for segments of holdout consumers for whom only characteristics are available, NMB procedures perform better than MB procedures. However, if it is important to understand the true segmentation structure in a market as well as the nature of the regression relationships within segments, the MB procedure is clearly preferred. Bicriterion segmentation methods are shown to be advantageous when there is at least some concordance between segments derived from different bases. Insights from the simulation study shed new light on a social marketing application in the area of segmenting and profiling overweight youths.