Research And Design Of Broadcast Broadband Customer Churn Warning System

With the increasing popularity of broadband Internet access,operators of broadband users development are also increasingly fierce competition,and for the broadband access service has become the focus of the loss of the operators,in the maintenance of existing customer relationships,not accurate and timely to face broadband customer churn capture and behavior analysis.Therefore,how to improve the accuracy of the prediction of customer churn will become a very important activity for all operators.This paper will mainly research on Data Mining Algorithm in the broadcasting industry in the broadband churn problem,proposed by simultaneously on several data model calculations were compared with the actual situation,confirmed by the evaluation and optimization of the only model algorithm,the prediction results of the final extraction scheme.The main work is as follows:1?Study the structure system and process model of data mining algorithms.Through the current mainstream CRISP-DM process model as the basic framework of this paper,it provides the theoretical basis for the analysis of customer chrn prediction methods.2?According to the characteristics of the radio and TV broadband business,this paper puts forward the design scheme of customer churn early-warning system based on data mining algorithm,and makes a comparative analysis of the decision tree,neural network,Logistic and other models.3?Through the data cleaning and extraction,the model variable selection,the structure of the derived variables,the actual user data,the calculated results are compared with the actual data,the final completion of the system test.The test results show that the decision tree model is the most suitable for the study of the analysis algorithm,and greatly improve the accuracy of user prediction results.Through the use of data mining project in some cities,provide regular list churn warning to the business department and customer service department,the Department will be based on the loss probability analysis and the actual situation,to stay flexible,improve work efficiency,reduce the risk of loss.So as to effectively control the risk of loss of broadband users,providing an important guarantee for the development of integrated services.