Research On The Telecom Enterprise Risk Early Warning Based On Rough Set And BP Neural Network

With several fatal reforms in China’s telecom industry, The competition ofvarious telecom operators is becoming more and more fierce, it focuses on customerresource competition, having more customer resource is the key to the operators towin in the competition, but in the telecommunications industry, When thesecompanies competed for a new group of customers, there will be accompanied by theloss of old customers. In our country, this issue has caused wide special attention andbecome the focus of operators, and it is also a problem that major operators are eagerto solve, this request predict the trend of customer churn before customer churn, andtake some reasonable measures to retain it.At first, this paper introduced the present situation of the telecom industry, andanalyzed the concept and characteristics of customer churn risk, introduces thetheory of rough set and BP neural network, laid a solid theoretical foundation for builttelecom customer churn enterprise risk early warning index system.Secondly, using the survey results, analyzed the causes and factors of the telecomcustomer churn, through reviewing and studying litertures, summed up the principlesof establishing risk early-warning index system of customer churn and methods,through actual investigation and interviews, using rough set attribute reductionalgorithm established telecom enterprise risk early warning index system of customerchurn.Then, through the analysis of the characteristics of BP neural network algorithmand modeling steps, built risk early warning of the BP network model which issuitable for the telecom customer churn enterprise, using collect customer churn riskindex data from the telecom companies for training and simulating the network model,proved the usability of BP network.Finally, the constructed BP network model was applied to the China Mobilebranch M, output warning result, and formulate relevant measures to improve theretention rate of customer according to the result of early warning.