Enterprise Financial Distress Prediction Based on Machine Learning and SHAP Interpretability Analysis

Enterprise financial distress prediction is a crucial aspect of financial risk management. Accurately identifying potential financial risks is of significant importance for investors and financial institutions. This study utilizes a dataset containing 3,672 samples and 83 financial features to build multiple machine learning models for predicting corporate financial distress. SHAP (SHapley Additive exPlanations) is applied to analyze the interpretability of model decisions. A complete workflow is established, including data preprocessing, feature selection, model training, and evaluation. Correlation analysis is used to identify key features most associated with financial distress. Among the algorithms tested—such as decision tree, gradient boosting, and logistic regression—XGBoost achieves the best performance with an F1-Score of 0.242 and a ROC-AUC of 0.910. SHAP analysis reveals that features like x36, x44, and x13 play crucial roles in prediction, offering a reliable tool and decision support for enterprise financial risk assessment. In particular, this interpretable framework provides a scalable and cost-effective solution for small and medium-sized enterprises (SMEs), which often lack access to advanced risk management systems. By enhancing early warning capabilities and decision-making transparency, the proposed method supports operational resilience and long-term financial sustainability for vulnerable