Data Mining In Telecom Customer Churn Prediction

Along with the development of the market competition in telecom industry, the telecom operators contend for customers more and more drastically. Data mining techniques is a good solution to prevent and reduce Chum. By employing data mining technology, we can: First, build a prediction model for Chum; Then, use the model to analyze why customers Churn and which customers are most likely to Churn in the future; Finally, make e make better target recruitment campaigns by summarizing customer's calling behavior and hobby to increase retention. So the operators can take action before Chum to reduce or avoid Chum.This paper aims for providing decision-making support for telecom industry in solving Chum problem and researches on how to employing data mining techniques to build a prediction model for Chum in the base of data warehouse and data mining techniques. The main works are as follows: Customer's classification by clustering algorithm. The purpose of customer's classification is to get different cluster which has common calling behavior, and then the prediction model will be built based on these different clusters. A modified k-meansmethod which can reduce compute complexity greatly is adopted to cluster similar customers; Chum prediction by decision tree classification. Classification is one of the important methods for Chum prediction. Churn prediction adopts decision trees algorithm. In the base of the basic algorithm, we also research on pruning and attribute selection measurement; Design and implementation of a Chum prediction system. First a Churn prediction system Architecture is proposed. Then By the model establishing process which includes Business Understanding, Business Understanding, Data Choice, Data Preparation, Modeling and Evaluation, the more accurate model is built for providing decision-making support for telecom operators.Statistic Analyse Management System is on work now. The system one side changes the work mode, the other side improves the veracity of prediction. Now, using data mining to support decision-making for the telecom operator is still in the phase of exploring. We hope that the work in this paper can do a little contribute to advance applying data mining for telecom industry.