This study addresses the critical need for accurate and explainable credit default prediction models, given the significant financial and regulatory implications of loan defaults. Despite advancements in machine learning, traditional methods like Logistic Regression remain favored due to their interpretability. The research employs several tree-based ensemble methods, with XGBoost being the most effective model. The study uses SHapley Additive exPlanations (SHAP) to enhance the explainability of the ML-based credit scoring models, using data from the US-based P2P Lending Platform, Lending Club. The results show that while all models perform similarly, XGBoost outperforms others in terms of accuracy, precision, recall, and AUC. The study also highlights the challenges posed by the imbalanced nature of credit default data, affecting precision values. By meeting a predefined standard of explainability, the proposed model holds promise for practical industrial deployment, addressing regulatory requirements and ethical concerns. Future work could further refine and validate the methods on diverse datasets.This study addresses the critical need for accurate and explainable credit default prediction models, given the significant financial and regulatory implications of loan defaults. Despite advancements in machine learning, traditional methods like Logistic Regression remain favored due to their interpretability. The research employs several tree-based ensemble methods, with XGBoost being the most effective model. The study uses SHapley Additive exPlanations (SHAP) to enhance the explainability of the ML-based credit scoring models, using data from the US-based P2P Lending Platform, Lending Club. The results show that while all models perform similarly, XGBoost outperforms others in terms of accuracy, precision, recall, and AUC. The study also highlights the challenges posed by the imbalanced nature of credit default data, affecting precision values. By meeting a predefined standard of explainability, the proposed model holds promise for practical industrial deployment, addressing regulatory requirements and ethical concerns. Future work could further refine and validate the methods on diverse datasets.