The Study and Realization of Customer-Churn Model Based on Date Mining in Telcom

Customer frequent churn is a serious problem in telecom. In the three major telecom operators, the competition is quite fierce. Owing to lack of a high-efficient prediction model ,the existing means effect is far from enterprise target. This paper proposes a combination model CPM based on constraint model, prediction model and mark model responsible for different job. Customer subdivision is vital for pertinent service further to reduce the rate of latent customers run off.