An explainable financial risk early warning model based on the DS-XGBoost model

Abstract

This study constructed a financial risk early warning model based on the d-S Evidence theory-XGBoost (DS-XGBoost) and analysed the model explainability combined with SHAP. Taking China's listed manufacturing companies from 2012 to 2021 as samples, this paper combines financial indicators, corporate governance and management perception to construct a more comprehensive financial risk early warning system. It is found that the model constructed in this study improves the performance of the financial risk early warning model, quantifies the contribution and correlation of features, enhances the model's explainability and reliability, and can provide information users with a more accurate decision-making basis.