Telecom Customer Churn Prediction And Retaining Measures Research

With the development of the international telecommunication industry, mobilecommunication industry has entered the era of3G. Thanks to the Chinese economicreform, China’s mobile communications industry developed rapidly. In order to meetthe domestic demand, the restructuring of China Telecom Industry have completed in2008. Recombination allows three companies "full service" operation After thereorganization of the formation of the three mobile communication operators: ChinaMobile, China Telecom, China Unicom. It also makes the domestic mobilecommunications market competition increase. A is a regional branch of China mobile.Because of the3G network, star terminal and other reasons, A lost a part of marketshare and revenue. In order to maintain and enhance market share and revenue, on theone hand they take measures to scramble for new users, on the other hand, they hope tomaintain existing customers (i.e. reduce "off-grid" user). Therefore, in order to help Ato reduce "off-grid" user and enhance market returns, based on data mining knowledge,we establish prediction model and find out the target users, then according to thecustomer relationship management and marketing theory, we developed customerretention program.We use C4.5decision tree method modeling, the data comes from A’s actual userdata, according to the alert level from high to low, we will divided object into red,yellow and blue warning user. In this paper, the model is evaluated with real data, theaccuracy rate of model forecast of47.3%, coverage rate is43.3%, the predictionaccuracy of red warning the user is77.2%, and coverage rate is54.7%. Based on theprediction results, we proposed some customer retention schemes, it includes:(1)customer classification,(2) layered retention measures,(3) the establishment oflong-term customer retention system. The results shows that the prediction model andcustomer retention schemes have some effect.