Hybrid Credit Risk Evaluation for Supply Chain Finance Based on Machine Learning and Traditional Risk Indicators

Abstract

Credit risk evaluation in supply-chain finance requires the integration of enterprise financial indicators, operational relationships, market conditions, and repayment behavior. Traditional credit scoring models often rely heavily on static financial ratios, while machine learning models can capture nonlinear risk patterns but may lack interpretability. This article develops a hybrid credit risk evaluation framework that combines traditional credit risk indicators with machine learning algorithms, including support vector machines, gradient boosting, and genetic algorithm optimization. The framework is designed for supply-chain finance environments where small and medium-sized enterprises are affected by core-enterprise stability, transaction authenticity, logistics reliability, and industry volatility. Explainable AI techniques are incorporated to clarify the contribution of financial and non-financial variables to default predictions. The proposed model provides a structured method for banks, platform enterprises, and financial institutions to assess credit risk more accurately while maintaining transparency. The article highlights the value of combining domain knowledge with algorithmic learning to improve credit risk warning and decision support.