Credit Risk Assessment using Ensemble Models and Explainable AI

New challenges are still there in the credit risk management in this context of the financial system, primarily because of the lack of good models for prognosis as well as low interpretability. To address this issue, this research brings together CatBoost, XGBoost, and LightGBM to improve the prospect of identifying high-risk borrowers. A minimal level of Explainable AI (XAI) is utilized to make confirmations that the model follows them and to build confidence in the outcome. Moreover, based on a difference to basic models, the obtained ROC-AUC score equals 0.9693, proving the feasibility of the proposed ensemble approach in questions of impact developing default risks. The end product is therefore a robust method with ensemble modeling and compliant with regulatory requirements in operational financial environments.