Building And Analyzing Default Probability Model Based On Logistic Regression

As the commercialization reform of domestic bank has just started,laws、 regulations、data and management methods are not perfect,which produce fi-nancial risk.Credit risk is the main risk of commercial Banks in China. With China’s accession to the WTO.the financial market of our country further opening to the outside world.The currently, the main task of China commer-cial bank face to is that how to take effective measures to connect with in-ternational banking,which bear the brunt of how to manage credit risk.The implementation of Basel2require Banks to establish a complete internal cred-it rating system, in order to taking credit rating of customers and quantization of loan customers credit risk.In the paper, through the establishment of the default probability model to predict customer defaults, to further strengthen the credit risk management of commercial Banks.In the paper,default probability model is based on Logistic regression anal-ysis.Firstly,using back stepwisc to select index of model.secondly,through the Logistic regression to get default probability model.At last, We use Kolmogorov-Smirnov test and ROC curve to test the model’s ability and accuracy of distin-guishing default customer and to get default probability model. A commercial bank through an empirical analysis of the data,draw conclusions.