Segmenting Financial Services Market: An Empirical Study of Statistical and Non-parametric Methods

Abstract

In this paper, we analyze segmentation of financial markets based on the general segmentation bases. In particular, we identify potentially attractive market segments for financial services using a customer dataset. We develop a multi-group discriminant model to classify the customers into three ordinal classes: prime customers, highly valued customers, and price shoppers based on their income, loan activity, and demographics (age). The multi-group classification of customer segments uses both classical statistical techniques and a mathematical programming formulation. For this study we use the characteristics of a real dataset to simulate multiple datasets of customer characteristics. The results of our experiments show that the mathematical programming model in many case consistently outperforms standard statistical approaches in attaining lower Apparent Error Rates (APER) for 100 replications in both high and low correlation cases.