Research On Activation Rate Of Credit Card Based On A Commercial Bank

In recent years,with the transformation and upgrading of China's economy,the development of consumer finance industry is more and more rapid.Credit card has become the main source of bank income.With the rapid development of credit card business in China,banks are increasing the number of credit card issuance and introducing various kinds of preferential cards,in order to maintain a certain market share and receive higher revenues.So it leads to the phenomenon of”one person,many cards",that is,a cardholder owns many cards,but does not activate therm,which not only causes the low activation rate of credit cards,but also brings unnecessary loss to banks.Throughout today's domestic credit card market,the fierce competition among banks is intensifying.The functions and benefits of most kinds of credit card are similar.In addition,the death rate of credit card is still high.Blind market selling model is outdated.Banks urgently need to introduce advanced Internet technology to trap the real needs of customers.Based on the above phenomena,this paper establishes a new card activation model from the bank's point of view.Through the analysis of customer information,it predicts whether the customer will activate the credit card in the future.On the one hand,it can reduce the pass rate of the customers with low activation rate,on the other hand,we can introduce and sell more products to the customers who have high activation probability,improve the overall card activation rate and reduce unnecessary loss.Since there are few studies on new card activation models at home and abroad,this paper refers to credit card scoring model to establish activation model.The main work of this article is as following:1.By introducing the development history of credit card,analyzing the current situation of domestic credit card market,the application steps of applicants and the process of bank auditing,this paper leads to the necessity and practical significance of establishing the activation model of new credit card,as well as the specific steps that can be applied in banks after the establishment of new credit card model.2.Based on the actual needs of banks,six specific steps are proposed to build the new card model.Including:data acquisition,exploratory analysis&data cleaning,variable boxes,variable selection,model architecture and model validation.For the selection of input variables,this paper considers both qualitative and quantitative variables:for qualitative variables,first,variables are subdivided into boxes,and WOE and IV values are calculated after subboxes;for quantitative variables,stepwise regression method and Lasso regression are used to select variables.3.For the specific structure of the model,this paper chooses 19980 real data of commercial bank credit card customers in a certain year of 2013.After desensitizing information processing,it includes basic information of customers,loan details,credit card details,pedestrian inquiries and so on.Using R,data missing value processing,index preliminary processing and variable box-dividing are carried out in turn.Based on 10-fold cross-validation,three statistical methods are used to establish models(principal component analysis + stepwise regression + logistic regression,lasso regression + logistic regression and lasso regression + principal component analysis ?logistic regression).The significance of model fitting is tested respectively.Gain and Lift graph are used to evaluate the model.Finally,the validation data are used to verify the model,and the best modeling method is obtained.The results show that the principal component analysis + stepwise regression + logical regression method has the best effect on the selection of variables and model fitting and prediction.This paper takes the data of a commercial bank as an example to establish a new card activation model,which has high application value in practice.It can predict the activation rate of customers in advance and devide the customer groups accordingly.Different marketing strategies and product customization schemes can be adopted for different kinds of customers to improve the overall activation rate and utilization rate and increase bank profits.