Credit Scorecard Model Based On Data Mining Technology

Due to market segmentation,intense competition among industry,huge customer behavior recorded,traditional credit audit has been unable to meet the bank and financial institutions to determine the customer's risk of credit and customer's rating.How to evaluate customer risk and realize the match between risk and profitability is a major problem for banks and financial institutions.In order to judge the risk of customers,achieve a balance between risk and profitability,we set up the scorecard model.This article through the following five aspects to describe it.First of all,through the study of historical documents and earlier credit scorecard model to understand the development of the scorecard model and how to use data mining technology to its empirical.Secondly,the choice of variables.Combined with the data of the financial company,final variables are age,sex,marital status,education,salary,housing,company nature,whether own a car,whether own property.Third,using the principle of data mining and SAS,R software,the logistic regression,decision tree,SVM and neural network classification model are established and synthetically analyzed,it is considered that the logistic classification model is more suitable for the purpose of this paper.Through the analysis,customer's age,gender,housing,company nature,whether own cars have a certain impact on the customer's bad good ratio.Fourth,the variables are scored by Evidence Weight Transformation(WOE)to get a total score for each customer.Enterprises can combine their own business needs,the actual experience and the customer's total score,divid customers into various levels,so that they can achieve market segmentation,complete customer positioning,balance the risks and benefits.Finally,the paper is summarized.