Research on Financial Risk Assessment Algorithm Based on CatBoost and SVM

This article focuses on the application of machine learning algorithms in financial risk assessment, particularly comparing the performance of CatBoost and SVM (Support Vector Machine) models. To ensure the quality of the input data, we preprocessed the raw data, including normalization, normalization, and filtering important features through correlation coefficients. In the experimental section, MAE (Mean Absolute Error) and RMSE (Root Mean Square Error) were used as evaluation metrics, and the results showed that the CatBoost model achieved an accuracy of 91.1%, significantly better than SVM and other traditional methods. In addition, this article also explores the improvements of gradient boosting trees (such as CatBoost) compared to traditional GBDT, such as the application of ordered boosting strategies and various regularization techniques to reduce overfitting and improve model generalization ability. This study not only validates the effectiveness and advantages of advanced machine learning algorithms in financial risk assessment tasks, but also provides reference directions for subsequent algorithm optimization.