Comparing approaches for clustering mixed-mode data: an application in marketing research,

Practical applications often involve mixtures of categorical and continuousvariables. A variety of algorithms has been proposed for clustering mixed mode data.In this paper we apply different techniques on a marketing research problem. We alsopropose an approach based on the consensus between partitions obtained by consideringseparately each variable or subsets of variables having the same scale. This approach maybe applied to data with many categorical variables and does not impose restrictive assumptionson variable distribution. We then suggest a final fuzzy partition with membershipdegrees obtained as a function of the classes determined by the different methods