Multiple Criteria Quadratic Programming for Fund Customer Churn Analysis

Abstract

Customer churn analysis and prediction play an important role in customer relationship management and improve benefit of enterprises. In recent years, classification models based on mathematical programming have been widely applied to customer churn analysis and have been proven to be powerful tools. In this paper, a new Multiple Criteria Quadratic Programming (MCQP) model is proposed and tested using fund customer dataset. We use ten-fold cross validation to test the accuracy and stability of the model. Finally, we compare our model with other three well-known models: Decision Tree, Artificial Neural Networks and SVM. The results show that MCQP is accurate and stable for predicting the customer churn. Consequently, we can safely say that MCQP model is capable of providing stable and credible results in predicting customer churn.