Research On Customer Churn Prediction And Management Srategy Of CEB Lanzhou Branch Based On Data Mining

In recent years,with the development of social economy in our country,and the constant improvement of the financial system,the policy Banks,joint-stock commercial Banks,regional local Banks,foreign Banks sprang up all over the country,such as various types of bank even set branches in the community surrounding,the rapid development of the banking sector for the development of social economy provides a solid guarantee,but also intensified competition in the banking itself.At the same time,the rapid development of information technology and Internet technology,the Internet was born in the financial is to provide a fertile soil,along with the online banking,mobile banking,bank of WeChat,and a large number of the emergence of the Internet financial products made the financial concepts of regional business gradually blurred,regional differentiation gradually reduce,the customer's choice of financial services and financial products are increasingly free and diverse,customer dependence of banking institutions and loyalty is more and more low,the commercial bank customer churn problem increasingly highlighted.According to the above research background,this paper,based on the related theory has been separated into four parts.The first parts deeply discussed about the general customer relationship management and customer losing situation of China Everbright bank(CEB)Lanzhou branch,then applying data mining technology to establish CEB's prediction model,and analyze it‘s necessity and feasibility.Secondly,due to the data mining modeling process,this paper choose the target customer group to identify the customer losing concept,the basic data including the first and second season in 2017 of CEB Lanzhou branch,customer's assets and its changes data,their trading behavior,the above data are the sample data of model building.Then the chosen target variable and 215 initial model variable use relative test method,during this process to dislodge those useless model variable,the final passing 75 models have been chosen to build the customer losing prediction model,by using R language technology and the logistic regression model method to build customer losing prediction model,through testing the effort that data group affect on the model,the result shows this method can well distinguish the target customer,losing customers and exist customers have been well distinguished,the accurate rate reached at 91.5%,the effect is obviously achieved our goal in modeling.In the third parts,this article gives the detailed instruction on prediction model,Then apply the model to research CEB Lanzhou branch's actual customer losing prediction,and describe the whole work process of customer detainment in marketing.In the end,we summarize and forecast the working practice of the application forecasting model to retain marketing customers.This article adopts the method of combining theory and practice,Uses the advanced data mining technology,building customer churn prediction model and use the everbright bank lanzhou branch actual customer churn problem,not only provides a good theoretical guidance and practical experience to the lanzhou branch of everbright bank in the management of the customer,as well as provides certain practice reference to solve the problem of customer churn of other branches and other domestic commercial Banks.