Identifying Small Market Segments with Mixture Regression Models

Abstract

The purpose of this work is to determine howwell criteria designed to help the selection of theadequate number of market segments perform inrecovering small niche market segments, in mixtureregressions of normal data. As in real world data thetrue number of market segments is unknown, theresults of this study are based on experimental data.The simulation experiment compares 27 segmentretention criteria, comprising 14 information criteriaand 13 classification-based criteria. The results revealthat AIC3, AIC4, HQ, BIC, CAIC, ICLBIC andICOMPLBIC are the best criteria in recovering smallniche segments and encourage its use.