The Analysis And Design Of The Telecom Churn Model Based On Data Mining

With the rapid development of the technology of data warehouse and data mining, customer relationship management (CRM) becomes more and more important. On this need, the investigation of predictive system of telecom churn based on Data Mining is introduced in this paper. Some arithmetic of DM is combined by the Data Mining theory. Furthermore, a new kind of predictive system of telecom churn is created. This model can improve the classification precision and recognition efficiency effectively, and makes full use of the advantages of clustering and logistic regression. This method of analysis is to find the implied operation rules according to the model which is based on the known data, and then to make the prediction with the rules above to guide the decision-making.The main research of this paper is the application of Neural network, Decision Tree and Logistic regression in the analysis of telecom churn. Firstly, the theory and correlation arithmetic is introduced, and a detailed description of them is made. Secondly, based on the practical situation of telecom corporations, the importance of application of DM is analyzed, and the basic description of predictive system is given according to the practical requirement. Finally aiming at the problem of telecom customer churn a lot of historical data is analyzed by using technology of DM. And the efficiency and precision of three methods have been analyzed and contrasted by using the CRISP_DM (Cross-industry Process for Data Mining) frame and the step of business understanding, data understanding, data preparation, modeling evaluation and development. At last the model which is found by Logistic regression is chosen because of its good evaluation index. And the DM tool (Clementine) is used to complete the design and realization of predictive system of telecom customers churn. Based on the investigation above, a predictive system of telecom customer churn is built in this paper, together with the characteristics of the forecast system. The solution scheme of telecom churn is mentioned.