Strategies for consumer segmentation with applications on preference data

This paper presents various strategies for the identification of consumer segments. One issue is to employ some segmentation method on the "long-thin" versus "short-fat" orientation of the consumer preference data. One strategy is to assign consumers with no strong preference to one segment, and then apply fuzzy clustering (FC) in the segmentation on the remaining consumers. The segments are evaluated by ANOVA and from external preference mapping to sensory descriptive data. Shuffled preference data are subject to segmentation to investigate the robustness of the procedure. The results indicate that segmentation on the long-thin data gives a better relation to the sensory data, and that segmentation on the short-fat data without defining the non-reliable segment gives rise to spurious segments. (C) 2004 Elsevier Ltd. All rights reserved.