<p>Accurate prediction of corporate financial performance is crucial for investors, financial institutions, and policymakers. While machine learning (ML) models, particularly ensemble methods like XGBoost, have demonstrated superior predictive capabilities, their inherent “black-box” nature often hinders their adoption in finance, a domain emphasizing transparency and interpretability. This research addresses this challenge by integrating SHapley Additive exPlanations (SHAP) with ML models to not only predict corporate financial performance but also to provide comprehensible insights into the predictive drivers. A comprehensive framework is proposed, leveraging XGBoost for financial classification tasks, complemented by SHAP for in-depth interpretability. The research evaluates model performance using metrics such as Accuracy, Precision, Recall, F1-score, and AUC-ROC, based on financial data of publicly listed companies in Taiwan from 2018 to 2024. Furthermore, the study investigates the stability of SHAP-based explanations across different years, a critical aspect for reliable decision-making. Key findings indicate that XGBoost significantly outperforms other benchmark models. The SHAP analysis successfully identifies key financial indicators influencing performance predictions and reveals the temporal stability of these feature contributions. This research contributes by offering a robust and interpretable financial prediction framework, enhancing model transparency and trustworthiness for practical applications in corporate finance and risk management. The findings underscore the utility of combining advanced ML techniques with XAI for more reliable and understandable financial forecasting.</p>