Linear Scoring Model Of Credit Card Based On The Component Analysis Methods

Credit card, as a personal financial instrument, it has some inherent risks, such as credit risks, operating risk, fraud risk, management risk and so on. It is very important for the banks to control credit card risk, so as to reduce the loss of breach. And the methods of credit card scoring model are developing depending on the requirements.The main content of this article is the linear credit scoring model based on Principle Component Analysis (PCA) and the Kernel-PCA. The article is divided into five sections.First, the article shows the history of credit scoring model, summarizes the present situations at home, several common methods for credit scoring, and compares their advantages and disadvantages.Second, the article introduces the basic thought, theoretical results as well as detailed implementation steps of Simplex method, the PCA method and Kernel-PCA method, which providing theoretical foundation for the program of data processing and the establishment of linear model. Consid-ering the possibility of linear relationship between the variables, the article uses the PCA data extraction. Based on the main components value of ev-ery customers, it establishes the linear scoring model, that is, willing all the components as independent variables, and work out the weights of them. Then, if there is a new customer, it can assess weather it is "Good" or not. Considering the possibility of nonlinear relationship between all variables, the article puts forward the Kernel-PCA data extraction, and at the same time, establishing the model accordingly.Through empirical analysis, the article realizes the linear scoring model based on the PCA data extraction. Here, we give all the components some economic significance. Meanwhile, the article proves the existence of nonlin-car hypothesis between variables, as well as the feasibility and effectiveness of linear model based on Kernel-PCA.Finally, the article points out the work that needs to research for further, looking the developments of credit card at home into the distance.