Research On Customer Groups Churn Prediction Model And Application Based On Survival Analysis

Increasingly personalized market demand and intensified market competition make enterprises pay more and more attention on customer relationship management (CRM) that considers customer as the focus of business. The cost of acquiring a new customer is 5 to 6 times of keeping an old customer. A modest increase of customer retention may effectively improve corporate profit. Therefore, an important task of enterprise is to predict the tendency of customers churn and maintain valuable customers under with limited resources. Customer churn management gets more and more attention of the academia and the business. However, the research focus is mainly on the improvement of customer churn prediction accuracy and analysis on reasons of churn.Present study mainly aims at individual customers, making customer churn prediction and customer retention management research are relatively isolated and hard to realize effective combination. Setting up integrated customer churn management system would significantly improve the efficiency of customer churn management. Looking for customer churn rule of customer groups with different characteristics, and grasping the dynamic and the time of loss are important for intergrated management of customer churn prediction and retention. This is very helpful for reducing costs and increasing profits.This paper build a customer churn model based on survival analysis for customer groups with different characteristics. Specific research is carried out from two aspects: Study 1:Identify customer groups with different characteristics which are similar within the group based on the RFM classification. Build a customer pridiction model by using survival analysis method. A sample of customer transaction data of a shopping mall in Dalian is applied to explore the state and time of loss among different customer groups. This paper found that different customer groups are significant different in the state of dynamic loss and developed the theory of customer churn management. Study 2:Extend the research based on the customer groups churn prediction model. Make sure of the necessity of maintaining one customer group according based on customer lifetime value. This paper helps for the research of integrating customer churn prediction and keep.Finally, this paper summarizes the relevant research conclusions and innovation points, analyzes the deficiency of this study, and discussed the future research direction.