OPTI-Recourse: An Explainable Machine Learning Framework for Credit Risk Assessment

A new tool called OPTI Recourse helps assess who might struggle to repay loans. It blends strong prediction models with clear explanations of how decisions are made. Instead of just giving results, it shows why they happen using real world data patterns. At its core sits a modified XGBoost setup tuned for uneven financial records. Predictions turn into straightforward ratings anyone can grasp quickly. Hidden influences behind each outcome come to light through SHAP and LIME analysis. Clarity improves when people see what shaped a verdict. Financial teams gain insight not only into risks but also their roots. It turns out clear models can also predict well, which fits neatly into strict regulatory settings. Right now the emphasis sits firmly on getting predictions right and explanations understandable. Down the line, efforts will weave in fairer training methods along with ways for people to act on feedback. Trust grows when systems show their reasoning while staying ethically grounded. Progress here points quietly to smarter, open, usable AI tools in how lenders judge risk.

1.         Chen, Y., Calabrese, R., Martin-Barragan, B.: Interpretable machine learning for imbalanced credit scoring datasets. Eu- ropean Journal of Operational Research 312(1), 357–372 (2024). https://www.sciencedirect.com/science/article/pii/ S0377221723005088 2.         Ahmed, M.R., et al.: Enhancing transparency and fairness in automated credit decisions using hybrid machine learning. Scientific Reports 14(1), 75026 (2024). https://www.nature.com/articles/s41598-024-75026-8 3.         Kisten, K., et al.: Enhancing fairness in credit assessment: mitigation strategies and implementation. IEEE Access 12 (2024). https://ieeexplore.ieee.org/document/10420549 4.         Mestiri, S.: Credit scoring using machine learning and deep learning-based models. Data Science in Finance and Economics 4(2), 236–248 (2024). https://www.aimspress.com/article/doi/10.3934/DSFE.2024009 5.         Tavakoli, M., et al.: Multi-modal deep learning for credit rating prediction using financial statements and market data. Neuro- computing 585 (2025). https://www.sciencedirect.com/science/article/pii/S0925231225000821 6.         Li, J.: Area under the ROC curve has the most consistent evaluation for binary classification. PLoS ONE 19(12) (2024). https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0316019 7.         Kwon, S., Jang, J., Kim, C.: Credit scoring using multi-task Siamese neural networks. Expert Systems with Applications 255 (2025). https://www.sciencedirect.com/science/article/pii/S0957417425001234 8.         Lundberg, S.M., Lee, S.I.: A unified approach to interpreting model predictions. NeurIPS (2017). https://arxiv.org/abs/ 1705.07874 9.         Ribeiro, M.T., Singh, S., Guestrin, C.: Why should I trust you? Explaining the predictions of any classifier. KDD (2016). https://dl.acm.org/doi/10.1145/2939672.2939778 10.       Wachter, S., Mittelstadt, B., Russell, C.: Counterfactual explanations without opening the black box. Harvard Journal of Law & Technology (2017). https://arxiv.org/abs/1711.00399