Research On Customer Churn Prediction And Retain In The Mobile Communication Industry

In 2008, domestic operators finish restruction, forming Telecom, Mobile, Unicom three carriers. With the domestic market environment gradually mature, the communication business and product marketing model of three carriers are growing convergence. The pressure of competition is more and more high. High rate of new-customers has become the past. At the same time, high rate of customer churn has become a very common phenomenon. How to effectively predict the potential loss customers and improve the efficiency of customer retention, which has a very important role for the operators.This paper summarizes the domestic and foreign scholars’research results on customer churn.Our research on customer churn mainly from three aspects:the reason of customer churn, customer churn prediction and customer retention. And from the market environment, the operator itself, competitors, clients themselves four aspects to do the theoretical analysis.Also, we analyze the consumption data of off-net customers. Which attributes will have a big fluctuation and which attributes change smaller in consumption data of off-net customers,which will be good forattribute selection forcustomer churn prediction analysis.We mainly consider the customer’s own change on the customer attribute selection.By analysing the current consumption situation and the historical consumption range of the customer to predict customer churn. And the choice of properties and processing method need to adapt to the method of building a model. Based on a study sample of customer data which is from a mobile communication industry, through numerical processing and filtering prediction index, then we use logistic regression method to build customer churn warning model so as to predict the potential losing customers. Besides, we recommend retention strategies for these people from the case-base. The performance of customer churn prediction model is good, and it can accurately predict the loss of potential customers. It also explains the method by selecting forecast properties based on the analysis of the customer itself changes is correct. Case-base stores successful retention strategies, which can help improve the effectiveness of our work. The model of customer churn warning can make a good prediction for the potential losing customers, and through the case-base for worth to retain customers recommend suitable retention strategy.