Research Of The Model Of Customer Churn Prediction Base On E-Business

In recent years, with the development of the Internet and advanced information society, E-commerce website as a new marketing channel for enterprises, more and more consumers have started to accept this kind of online shopping purchasing way.E-commerce website has increasingly become a place of many Internet users visited at the same time also brought a series of problems, customers unstable, high customer churn rate. For the website, how best to avoid the loss of customers, reduce customer churn rate, has become one of the focus E-business operators should pay attention to.In this paper,firistly , based on discussing customer primarily theories such as CRM theory, E-business customer theory churn and DM theory, this dissertation, combining with characteristics of customer churn in E-business website ,four kinds of customer churn prediction model have been provided.Secondly, the study has implications for both theory and practice. Based on CDNOW website's customer data, using Pareto/NBD model, a Na?ve Bayes model, support vector machines model and BP arithmetic neural networks model to predict customer activity .In dissertation, a detailed description of the empirical process in Part IV including data discription and explanation, empirical model, model evaluation by numerical indicators .Finall, the cassification of website customer churn and principles of controlling customers churn are proposed on CRM theory and proofed results. Then introduce the idea of the forces on customers churn, analyzing the forces which effect website customer churn, discovery the thrust forces and gravitation forces.Finally, by analyzing, eontrolled retaining strategies and resistance strategies to control website customers churn From practice aspect, the proposed research ideas can provide a reference for our E-commerce industry about analyzing customers churn and prediction .