Research On Commercial Banks Individual Customer Churn Retention Based On Data Mining

In recent years,commercial banks are facing more and more fierce competition with accelerating the process of interest rates marketizing,rising the Internet financial and strengthening financial supervision,especially for the customers' competiton.Customers become a vital strategic resources for commercial banks with the buyer 's market developing.However,the customer churn of commercial banks are more and more serious at this stage,which becomes an important problem that bounds the development of banks.A research shows that,retaining an old customer is more valuable than getting a new customer for banks.As the limited resources of banks,the different value of customers and the different possibility of retaining customers successfully,it is unrealistic and wasteful to retain all customers.Therefore,how to evaluate the churning customers' value and predict the possibility of retaining successfully,find out customers who need to be retained,formulate reasonable retaining strategies,become a problem to be solved.This paper first comprehensively reviews and analyzes the domestic and foreign scholars research achievements about customer churn retaining.Based on this,the research content,the research methods and the research route is proposed.Secondly,through learning the customer churn retaining theory,the customer value theory and data mining theory,the paper redefines the concept of customer churn and the concept of customer retention,elaborates evaluation method for customer value and the process of data mining,and introduces briefly three kinds of data mining algorithms used in this paper.Then,based on the data mining theory and the customer value theory,the paper establishes the segmentation model of customer value and the prediction model of customer retention from customer value and whether to retain successfully.And combining the result of these two models,the paper determines customers who need to be retained.In the process of establishing the model of customer value segmentation,considering the influence of different value,the paper builds an index system of customer value evaluation about the commercial bank from the current value and the potential value.And on this basis,use the fuzzy c-means clustering to classify and get four customer groups with different value,so as to determine the value type of each customer.In the process of building the model of customer retention prediction,a retention prediction model based on rough sets and the support vector machine is proposed.Considering the problem that attributes having repeating information,use rough sets to reduce the attributes in this paper.And then,considering unbalanced data sets,this paper builds a prediction model based on the different price of misclassification,which uses the reduced attributes as the input and uses whether to retain successfully as the output.Based on this,use the cross validation method to choose the parameters of the model.Furthermore,do the empirical study for the proposed model combined with the specific conditions of Fujian branch of the A commercial bank,which found that the model can effectively identify the different customer groups and predict the retained customers.Through analyzing the customers' characteristics of the different value classes,the paper makes applicable retention strategies for the A bank.