The Application ofAdaBoost in Customer Churn Prediction

Since attracting new customers is known to be more expensive, the enhancement of existing relationships is of pivotal importance to companies. Therefore, as part of the customer relationship management (CRM) strategy, predicting customer churn and improving customer retention have attracted more and more attention. Being aware of the defection prone customers beforehand, companies could react in time to prevent the churn by offering the right set of products, modifying the sales strategy and providing customized services. Therefore, high predictive performance could ultimately lead to profit increasing for companies. In this paper, we use the AdaBoost which is a main branch of boosting algorithms to predict the customer churn. We have implemented three different boosting schemes: Real AdaBoost, Gentle AdaBoost and Modest AdaBoost. Applied to a credit debt customer database of an anonymous commercial bank in China, they are proven to significantly improve prediction accuracy comparing with other algorithms, like SVM. The assessment and comparison of these algorithms are made to analyze the traits of them. Data processing and sampling scheme are also detailed in this paper.