The Research Of Credit Scoring Model Based On Adaptive Lasso

With the improvement and development of China's financial market,small and medium sized enterprises have injected vitality into the market?Simultaneously,they also inevitably bring certain risks as well?When measuring corporate credit risk,there are often existing high dimensional variable indicators?Based on the Logistic regression model with good properties in the study of credit risk,this paper applies effective variable selection methods to build a credit scoring model with Logistic regression to make measurement analysis and forecast of China's SMEs' credit risk?This paper applies the Lasso variable selection and the adaptive Lasso variable selection with ridge estimation to the Logistic model respectively,which could be used to the SMEs' credit risk scoring?Thus variable selection and parameter estimation can be implemented at the same time?Next,in order to improve the model,the information value of variables are used to build the penalty weights in the Adaptive Lasso method,then this paper verifies its validity?This paper introduces the above methods and related algorithms,and conducts empirical research on small and medium-sized board companies listed in China?Due to the lack of information data in China's SMEs,the data used in the research is unbalanced data?This paper adopts the SMOTE method to preprocess the data,building a corporate credit index system and verifying the classification performance of each model?The research indicates that the Adaptive Lasso improved by the information value is superior to the ALasso method and the Lasso method?The selected variables are more representative and more comprehensive,and they have better performance in risk differentiation and classification prediction?It can also reduce the influence of uncertainties and improve the stability of the model?...