Enhancing Profitability Prediction in Vietnamese Commercial Banks With a Transparent and Explainable AI Approach

Predicting profitability in the banking sector is essential for effective financial management, risk mitigation, and sustainable growth. In the context of Vietnamese commercial banks, profitability prediction poses significant challenges due to complex relationships among financial metrics and the volatile economic environment. Traditional econometric models often fall short in capturing these complexities, resulting in suboptimal predictions. Although advanced machine learning techniques like XGBoost have shown superior predictive accuracy, they lack the necessary transparency and interpretability, making them challenging to deploy in practice due to regulatory concerns surrounding “black box” models. This study addresses these limitations by proposing an explainable machine learning framework that integrates XGBoost with SHAP (SHapley Additive exPlanations) to predict profitability. The framework uses key financial metrics, including Return on Equity (ROE), Return on Assets (ROA), Total Assets, and Non-performing Loans.