The Mobile Customers Churn Prediction System’s Design And Implementation Based On Data Mining

With the reform and restructuring of China’s telecommunications industry, China’s telecommunications market environment has undergone unprecedented changes between the increasingly fierce competition in the industry, scrambling to compete for customers. In such a customer-centric competitive environment, mobile operators as an important means if we want to cultivate the core competitiveness of enterprise services, customers must identify consumer characteristics, consumption patterns from the complex historical data and customer master consumer characteristics, consumer law is a top priority, therefore, customer segmentation work is an important foundation of customer relationship management through customer segmentation, and provide customers with personalized, targeted, such as differentiated services to step by step to improve the competitiveness of enterprises.Currently, how to predict customer churn and take appropriate measures to retain, has become a mobile operator in the customer relationship management urgent problems. But in the face of customer information mobile operators who have hundreds of millions, and every customer use, the system will generate a new message is recorded, you want to find that they are useful from such an enormous database information, that is a very difficult job. Faced with this dilemma, data mining techniques to solve the problem is undoubtedly the most effective way. Through data mining technology can effectively predict the propensity of customers to the existence of the loss, which can formulate corresponding measures to retain customers, so you can avoid a lot of churn caused by the loss of business profits and corporate image of the economic loss, Meanwhile, seriously affected the favorable position in the telecommunications market competition.In this thesis,using data mining techniques herein a series of movement data analysis, and data according to the characteristics of mobile data preprocessing established data set used herein to mobile data mining analysis. Combined with mobile business needs, data mining technology in the field of mobile data analysis.Accordance with the steps: ’Data Understandingâ†'Data Preparationâ†' Modelingâ†' Assessment Modelâ†'Publishing Model’, and use Microsoft SQL Server 2005 as a tool to build a decision tree algorithm using mobile customer churn prediction model. In the modeling process, carried out data cleansing, conversion, processing of imbalanced data sets, and the establishment of a churn prediction indicators. And the establishment of a decision tree model was evaluated for excavation to verify the accuracy and eventually come to a dominant factor in the loss of customers, loss of customers to keep the work provided.This thesis established a churn prediction model based on data mining of mobile clients, the results showed that the application of the established model can provide valuable predictive information to mobile decision makers and the corresponding solutions.