COMPARING SELF ORGANIZING MAPS WITH K-MEANS CLUSTERING: AN APPLICATION TO CUSTOMER PROFILING

Due to the increasing competitiveness in shopping environments, profiling consumers becomes critical for retailers and their managers. however, researchers investigated the consumer profiling based on either the shopping motivations, or the shopping values, or the consumers’ decision making styles separately mostly by using k-means clustering method in the literature. in this sense, a research investigating the shopping motivations and values, and consumers’ decision making styles together with a new clustering method, which overcomes the limitations of the k-means clustering method, is warranted in the literature. in this study, we used an artificial intelligence based technique, called self organizing map (som), to profile consumers based upon their shopping motivations and values, and decision making styles. our results also demonstrated that som’s total within cluster variance is smaller than k-means’s, indicating that som clustering is better than the k-means when the sample is non-normally distributed. the paper profiles clusters on demographics and ethnic group membership to examine similarities and differences among cluster members.