Analysis And Design Of Broad And Customer Churn Early-Warning System Based On Data Mining

With the completion of the restructuring and 3G operator licenses issued, Telecommunications market is becoming increasingly fierce competition, Three operators from the traditional single mobile or fixed business operations into a full-service operations, The broadband business as a key business as full service carriers are an important competitive positions. However, with the increasingly competitive broadband services, broadband churn a growing problem. How to discover and hold on to the loss of broadband users becomes a problem that operators need to solve.Based on the CRISP-DM, through Personnel exchanges with business. Using data integration platform to extract mass properties and behavior of broadband users attribute data from the EDW system and CRM, billing and other business systems. Based on a large number of user data, the first step is to build wide tables for broadband users, using artificial construct derived variable methods to achieve improved properties of broadband users; Using SPSS Clementine data mining tools to review the sample data for the broadband users; By association rule analysis and correlation coefficients to achieve the properties of the Statute of broadband users. Using Association Rules to fill the Vacancies value of broadband users.This paper solves the problem that the Non-equilibrium distribution of the sample. Select the decision tree C5.0 algorithm for broadband churn modeling. The next step is to using the sample data to test the model. Finally, releasing model that used in front-line operational staff and marketing.Using data mining techniques, Front-line operational staff for the provision of broadband users per month loss of early-warning list. Front-line operational staff can better targeted marketing, and then achieved significant reduction in the wastage rate of broadband. A city to solve serious problems of broadband customers, but also validate the data mining technology based on the loss of broadband customers warning is valid. This paper not only solves the serious problems of broadband users, but also validate the data mining technology based on the loss of broadband customers warning is valid.