Classification Of Credit Card-Forecast And Analysis For The Default

More and more people use credit cards upon its convenience nowadays.Whereas credit card can overdraft amount,and bank will charge a certain amount of interest due to the refund delay.In order to occupy the market,banks absorb customers as far as possible.Since there are some customers maybe default,so the bank should control the default risk.In this paper,a data set of a bank in Taiwan are exploited to predict whether a customer will default or not,models are established by using kinds of classification methods.Then in order to cut loss,some restrictions are imposed on the customers who will default by detection.The following methods are discussed:Naive Bayesian Model,K-Nearest Neighbor,Weighted K-Nearest Neighbors,Support Vector Machine,bagging,AdaBoost,Classification Trees,Random Forest,Neural Network.The comparison of the various methods is explored to select the most suitable classification model.Also in order to reduce the cost of supervision for the default customers,the most important variables are screened to reduce the complexity of the model.