Research On Forecasting System Of Telecommunication Client Re - Entry Network Based On Data Mining

Under the circumstance of Market Economy, in order to improve the market competitiveness, telecommunication operators carried out various attractive marketing activities. Along with the increasing of new customers, there was a phenomenon which is called "Customer Rejoin", that lead customers to be "Big In Big Out". And a large Rejoin customers not only caused a number of resources waste, also increased the marketing costs of the operators, how to curb the occurrence of this phenomenon has become the top problem to each operator. The purpose of this research is to predict the customers which is possible to rejoin the net, and take effective market strategies in order to reduce the possibility.Traditional rejoin identification technology is the basic information identification、 IMEI recognition and social compatibility recognition, these technicals can be used independently or intersectionally, but the recognition have low accuracy, long recognition time, and occupy a larger data warehouse resources, which can’t play a better prediction effect.With the development of data mining technology, there are more and more prediction research which based on data mining. This paper adopted the basic informations and behavior datas in a telecommunication operator of a province group company, use the CRISP-DM methodology framework for modeling process, exert C5.0decision tree algorith、C&RT algorithm and neural network algorithm to model and evaluate the results of various algorithms, then obtain the index weight system of the optimal prediction model, establish the telecom customers’rejoin model. Finally, adopting the JAVA programming language, using the MyEclipse enterprise work platform, taking MySQL Server as the database and Tomcat as the Web Server, following the MVC pattern, we use SSH framework to develop a telecom customer rejoin prediction system which based on B/S architecture.The system provides a powerful reference for identifing the rejoin customers at an early time and making the corresponding decision and deployment for telecom marketing analysts, meanwhile, it has a broad application prospect and practical reference value for reducing the cost of the enterprises in cards and increasing the customer retention rate.