The Implementation Of Data Mining In Telecom Churn Model

With the progress of data mining technology, the importance of data mining is approved by more and more person. It makes use of passed data to find out the underling business rule by the way of the establishing mathematics model. In other countries, many fields have successful applications with the data mining. In our country, with the focus of data mining, data mining's application and research will be wider. The prediction of customer churn in telecommunication is a bit hot.The main research of this paper is the application of Decision Tree, Neural Network and Logistic Regression in the analysis of telecom churn. Firstly, based on the practical situation of telecom corporations, the importance of application of Data Mining is analyzed. Secondly, the theory and correlation arithmetic is introduced, and a detailed description of them is made. Thirdly, there is a simple introduction of DM Software-SPSS Clementine, which is used in this paper. Finally aiming at the problem of telecom customer churn, one telecom company's data is analyzed by using the CRISP_DM (Cross-industry Process for Data Mining) frame, with the steps of business understanding, data understanding, data preparation, modeling evaluation and development. And the efficiency and precision of three methods have been analyzed and contrasted. At last the best is chosen to complete the design and realization of predictive system of telecom customer churn because of its good evaluation index. Together with the characteristics of the forecast system, the solution scheme of telecom churn is mentioned.The project is combined with theory of DM. The final project is applied to predict customer churn. The result indicates that the forecasting model accords with the practical situation scientifically and can afford the predictive information and the solution project to decision-maker. This predictive model is of significance in solving the problem of predictive telecom churn.