Applications Of Data Mining In Customer Churn Management

With the deepening of economic globalization, the competition between industries isbecoming increasingly fierce, especially in the telecommunications industry. In theinformation age, companies have a lot of data, in order to seize the opportunity toenhance the core competitiveness of enterprises, companies must explore the usefulinformation from large amounts of data. Customers are the source of company·s profits,especially enterprises of telecommunications, if we apply data mining ideas andtechniques to customer relationship management, there will be much benefits, not onlycan provide a scientific basis to quantify the enterprise management and decision-making,also can help more efficient use of limited company·s resources, and improve profits.Any companies are faced with customer churn prediction and management problems,customers churn let enterprises realize profits of the period extend and suffer huge losses.Currently telecom companies generally use providing personalized service, analysis ofcustomer satisfaction and loyalty and other methods for customer churn management, butthese methods are empirical approach depends largely on the clerk’s subjective awareness,its effectiveness and scientific are difficult to verify, does not fundamentally solve theproblem faced by enterprises.In this paper, focusing on the customer churn problem in a telecom enterprise in ACity, doing research and analysis of data mining technology in the telecommunicationscustomer churn early warning applications. Using R software build customer churnprediction model, and propose new ideas for telecommunications companies churnwarning. Paper achievements as follows:（1）Propose the algorithms which commonly used in customer churn predictionmodel basing on the basic research, analysis of customer churn early warning modelingprinciple.（2）Realize and evaluate the core part of this article-customer churn prediction model. Use R software build customers churn warning model, such as K-means cluster,Logistic regression model, decision tree model, associated rule model. By compare thedifferent four models and make assessment between them, and finally select the optimalmodel. Providing more effective customer churn early warning methods for enterprises.（3）Using powerful data mining and visualization tools R software, gives the codesin this article which all the work achieved.