Classifications of Credit Cardholder Behavior by Using Multiple Criteria Non-linear Programming

Abstract

Behavior analysis of credit cardholders is one of the main research topics in credit card portfolio management. Usually, the cardholder’s behavior, especially bankruptcy, is measured by a score of aggregate attributes that describe cardholder’s spending history. In the real-life practice, statistics and neural networks are the major players to calculate such a score system for prediction. Recently, various multiple criteria linear programming based classification methods have been explored for analyzing credit cardholders’ behavior. This paper proposes a multiple criteria non-linear programming (MCNP) approach to discovering the bankruptcy patterns of credit cardholders. A real-life credit database from a major US bank is used for empirical study on MCNP classification. Finally, the comparison of MCNP and other known classification methods is conducted to verify the validation of MCNP method.