Information gain (Information Gain, IG) is a feature selection measure commonly used in decision tree algorithms, which measures the predictive ability of a feature to the target variable. The IG algorithm evaluates the importance of features by calculating the degree of entropy reduction before and after dividing the dataset. Entropy is a measure of the purity of a data set. The greater the information gain, the stronger is the conditional entropy of the data set divided according to feature A. A larger IG value increases the influence of feature A on the classification. In practice, the IG algorithm can help us to screen the most influential features on the target variable from a large number of candidate features, reduce the complexity of the model and improve the accuracy of prediction. MATLAB simulation shows that the IG-SVM model predicts the credit risk of supply chain financing of SMEs under certain evaluation criteria The accuracy and accuracy of credit risk prediction are better than traditional forecasting methods.