Research on the construction of default risk evaluation model of supply chain finance based on machine learning

To improve the accuracy of default risk prediction in supply chain finance, a default risk assessment model for supply chain finance companies using machine learning technology is developed to improve the early warning ability of supply chain finance companies for credit risk. Data from multiple sources such as corporate finance, enterprise and market are combined, and the prediction accuracy of the company's credit risk is effectively improved through data cleaning, symptom detection and combining multiple algorithms. Experimental results show that the F1 value of the deep neural network (DNN) model reaches 89.1%, indicating a good prediction result, while the supervised random forest (RF) and SVM have their advantages in accuracy and recall. This model provides a new technical capability for risk management in supply chain finance and a practical theoretical basis for risk management in related fields.