Support Vector Clustering for Customer Segmentation on Mobile TV Service

Abstract

This article presents a proposal for customer segmentation through Support Vector Clustering, a technique that has been gaining attention in the academic literature due to the good results usually obtained. An application of this method to a sample from consumers in Brazil about mobile TV service is performed; we compare this approach with the classical hierarchical methods of cluster analysis. It is concluded that this methodology is effective in reducing the heterogeneity often present in customers’ data, improving the cluster segmentation analysis for managers.