Credit Risk Assessment of Small, Medium and Micro Enterprises Based on The Improved Logical Regression Model of Gradient Descent Method

Abstract

Due to the weak strength of small, medium and micro enterprises, banks need to judge the credit risk of the enterprise based on the company's past credit history and the company's invoice information, etc., and then establish a reasonable and effective quantitative credit risk model.First, through the analysis and preprocessing of the data, this paper extracts six indicators such as marketing profit margin and invoice invalidation rate, and comprehensively evaluates the credit risk of the enterprise from the three aspects of enterprise scale, supply and demand stability, and credibility, and establishes a corporate credit risk evaluation system. Then, a logical regression model is constructed to predict the probability of compliance of the enterprise, a loss function is formulated to describe the deviation between the prediction result and the classification result, and a stochastic gradient descent algorithm is used to obtain the optimal parameter value. Finally, the optimized logical regression model is applied to the credit risk assessment of 123 companies. The results show that the stronger the company's strength, the more stable the supply and demand relationship, and the higher the reputation level, the less likely its credit default risk is.