Consumer benefit segmentation using clusterwise linear regression

Abstract

A generalization of the Clusterwise Linear Regression method is proposed for benefit segmentation. The segments found comprise consumers that attribute the same importance to perceived product dimensions in the formation of preference. The method is particularly useful when consumers' evoked sets of products are small and collinearity hampers the estimation of preference models at the individual level. A FORTRAN computer program for clusterwise regression of preference data is described. The performance of the algorithm on synthetic data is investigated, and an application to data on elderly people's preferences for meat products is given. Special attention is paid to significance testing with Monte Carlo test procedures, and convergence to local optima.