Credit decision of SMEs based on Improved TOPSIS and decision tree

Abstract

Listen

Translate article

Objective: To study the credit decision-making of SMEs. Methods: ① evaluation model establishment: through AHP, PCA, decision tree and other methods, the improved TOPSIS credit risk quantitative evaluation model based on AHP, the enterprise loan demand model based on PCA, and the bank hierarchical risk prevention and control model were established to make decisions for the enterprises with credit rating. Then, it makes a quantitative analysis of the credit risk, makes a credit decision to maximize the bank's income, and gives the error and result analysis. First of all, through the crawler, we obtained 600 pieces of data about the average monthly sales, average monthly input, the strength of upstream and downstream enterprises, and the number of buyer returns. Secondly, the modified TOPSIS credit risk quantitative evaluation model is established to analyze the above five factors and the credit rating of the enterprise, and the probability of the enterprise's discredit is obtained. Then, the hierarchical risk prevention and control model is established, and the principal component analysis method is used to quantify the strength of the enterprise, so as to obtain the loan amount of the enterprise. Finally, the credit strategy to maximize the bank's income is formulated, and the bank's profit margin is 4.94%. ② Generalization training: Taking enterprises without credit rating as training samples, the decision tree and support vector machine are used for training and testing respectively, and the correct rate of decision tree is 90.44%, which is better than 81.59% of support vector machine. Secondly, combined with five kinds of data, the credit risk¹ quantitative evaluation model of TOPSIS is used to solve the credit failure probability of Unrated enterprises. Through the hierarchical risk prevention and control model, the loan interest rate of the enterprise is calculated, and the multiple nonlinear programming is transformed into linear programming for solution, so as to obtain the optimal credit strategy, which can make the annual profit rate of the bank reach 9.5%. Conclusion: after getting the five factors of small and medium-sized micro enterprises, we can judge the credit rating of enterprises through decision tree and make credit decision.