Credit Risk Assessment Of Small And Medium-sized Enterprises Based On Survival Analysis

SMEs are the backbone of China's economic and social development,but the production and operation of SMEs is facing a severe situation,and corporate financing is difficult to become the biggest stumbling block on the road to enterprise development.How to provide banks with a feasible credit assessment model for the credit risk assessment of SMEs has become a top priority.This article constructs a credit scoring system for SMEs that includes 20 internal environmental indicators and 10 macroeconomic indicators.The static logistic regression model and dynamic Cox model are used to study the impact of internal environmental indicators on corporate credit assessment.To further deepen the forecast of the Cox model Capacity,this article adds dynamic macroeconomic indicators to compare whether macroeconomic indicators will affect the accuracy of Cox's assessment of corporate credit.As the bank's SME loan data is not available,the SME loan data of the Dianrong platform is used for empirical research.The conclusion shows that the corporate credit assessment model established by the Cox model can better screen the companies that meet the requirements than other credit assessment models,and the overall forecast accuracy of the Cox model with macroeconomic indicators is improved.The probability of such errors is reduced.According to the data mining analysis,due to the lack of raw data,the accuracy of the results obtained without strict data processing is not high.Therefore,in order to evaluate the accuracy of the results,this article uses WEKA3.6 and The two softwares of ROSETTA perform various processing on the data.Use the artificial method to fill in or delete the data for the problem of missing data;use the synthetic minority oversampling technique to solve the imbalance problem of the decision attribute(whether default)data;use the rough set theory software Rosetta for Boolean Logical discreteness;in order to simplify the model without losing accuracy,Johnson algorithm is used to reduce attributes.The evaluation results show that the accuracy of the credit analysis results of the survival analysis Cox model is improved compared with the logistic regression model.The accuracy of the Cox model evaluation with macroeconomic factors is improved compared with the evaluation results of the models withoutmacroeconomic factors.The accuracy of the results of the Logistic regression model and the macro-micro Cox model are improved compared with the results of ordinary data processing.The survival Cox model has high accuracy and strong applicability.