Evaluation Framework and Characterization Based on XGBoost-SHAP Models

This study is dedicated to analyzing the key influencing factors in a multilevel assessment system using advanced machine learning algorithms, with a focus on introducing the XGBoost-SHAP model as the main analytical tool. XGBoost, a gradient boosting algorithm with excellent performance, provides an in-depth understanding of the importance of features through the detailed interpretation of the model's outputs by means of the SHAP method. In this paper, by constructing a multi-level evaluation system and combining the powerful analytical capabilities of the XGBoost-SHAP model, the prediction accuracy is improved, while providing clearer insights into the relationships between features and the interpretability of the model. The application of the model demonstrates its extensive potential in data analysis, model optimization, and decision support, providing strong theoretical and practical support for further exploration of data-driven decision-making processes.