Abstract

     In today’s Chinese economic system, market system, and development goals, MSMEs play an essential role. The availability of capital is a necessary condition for the development of MSMEs. However, in China, MSMEs still cannot effectively eliminate the development dilemma that financing is difficult and expensive. The deep-seated reason lies in the imperfect credit evaluation system of the banking industry for MSMEs. The purpose of this paper is to establish a mathematical model for credit risk evaluation of MSMEs and to establish a new bank credit evaluation strategy to support and back up MSMEs of different sizes accurately.

Introduction

     As economic globalization develops, Chinese MSMEs have seen rapid development. However, with the further development of MSMEs, the shortage of capital often becomes a significant factor limiting the further development of MSMEs. Due to the limitation of the size of MSMEs, their risk tolerance is weak, leading to further increase of enterprise credit risk and the difficulty of credit assessment by commercial banks. The purpose of this paper is to establish a mathematical model for assessing the credit risk of MSMEs and to establish new bank credit strategies to support and back up MSMEs of different sizes accurately. In this paper, an evaluation model is established by combining the machine learning model XGBoost algorithm with AHP, and the combined model is used to construct and analyze the model to quantify further and evaluate the credit risks of MSMEs. While ensuring the interests of commercial banks, more MSMEs are provided with loans to increase the amount of loanable capital of the whole society. Promote the further development of MSMEs and promote social progress.

Model Evaluation and Conclusions

     This paper establishes a reasonable mathematical model and formulates the bank’s credit strategy by evaluating and ranking the credit risk. We first use the analytic hierarchy process to determine the factors affecting the credit risk and then use the comparative method to quantify the importance of each factor. After the consistency test is passed, an acceptable pairwise comparison matrix is constructed. MATLAB software calculates the proportion of N factors in this layer in target Z. The proportion in the criterion layer is written into a vector and normalized to obtain the weight vector.

     Then, this paper uses SPSS software to quantify the stability of supply and demand in the criterion layer and then quantify the relationship between customer churn rate and reputation level given in the data to calculate the average value of the three levels of ABC. After data preparation, this paper classifies the eigenvalues for similar machine learning. We use Mpai software to substitute effective indicators into the machine learning model XGBOOST regression for training. In this paper, xgboost machine learning is used to substitute the normalized weight vector in the analytic hierarchy process to obtain the quantitative regression value of the company’s strength. Finally, the credit risk rating evaluation formula is constructed using the criteria layer weight and the lower level quantitative data. The final enterprise credit risk ranking is obtained using Mpai software, and the credit strategy is given scientifically and reasonably according to the ranking.