Ensemble learning for credit scoring: A comparative study of LightGBM and CatBoost

Do Thi Thanh Dieu, Ha Binh Minh, Phan Dinh Phung, Trinh Hoang Nam

DOI: 10.63065/ajafe.2026.1.004

Email: minhhb@hub.edu.vn

Đơn vị công tác:

Ngày nhận bài: 09/09/2026

Ngày nhận bài sửa: 09/09/2026

Ngày duyệt đăng: 09/09/2026

Lượt xem: 6

Downloads: 0

Các định dạng trích dẫn

Tóm tắt

Purpose – This paper examines whether gradient-boosted tree ensembles can improve credit-scoring performance over a traditional logistic model while remaining usable within common banking governance practices.

Design/methodology/approach – Using the UCI dataset (30,000 observations), we compare logistic regression, LightGBM, and CatBoost under consistent class weighting (w = 3.52). All models are tuned with Optuna (40 trials) using stratified 3-fold cross-validation. Evaluation includes ROC-AUC, PR-AUC, KS, and Brier metrics; DeLong’s test; SHAP interpretation; and calibration diagnostics via ECE and SSM regression. Gender fairness and PSI stability checks are also performed.

Findings – Both ensembles outperform logistic regression (ROC-AUC ≈ 0.777 vs 0.710; KS ≈ 0.42 vs 0.36). DeLong’s test confirms no significant difference between CatBoost and LightGBM (p = 0.768), though both exceed the baseline (p < 0.001). SHAP identifies repayment status (PAY_0) and credit limit (LIMIT_BAL) as dominant drivers. Post-hoc isotonic calibration improves LightGBM’s ECE from 0.200 to 0.008. Finally, ensembles reduce fairness gaps and maintain high stability (PSI < 0.01).

Originality/value – The paper offers a compact, reproducible evaluation workflow that integrates hyperparameter optimization, imbalance handling, statistical significance testing, explainability, fairness diagnostics, and stability monitoring for ensemble-based credit scoring.

Từ khóa

Calibration, CatBoost, Credit scoring, Ensemble learning, Fairness, LightGBM, SHAP, Stability

Download Mua ngay Buy now

Tài liệu tham khảo

Akiba, T., Sano, S., Yanase, T., Ohta, T., & Koyama, M. (2019). Optuna: A Next-generation Hyperparameter Optimization Framework. https://doi.org/10.48550/arXiv.1907.10902.

Badaro, G., Saeed, M., & Papotti, P. (2023). Transformers for Tabular Data Representation: A Survey of Models and Applications. Transactions of the Association for Computational Linguistics, 11, 227-249. https://doi.org/10.1162/tacl_a_00544.

Barocas, S., & Selbst, A. D. (2016). Big Data’s Disparate Impact. California Law Review, 104, 671-732.

Breiman, L. (2001). Random Forests. Machine Learning, 45, 5-32. https://doi.org/10.1023/A:1010933404324.

Brier, G. W. (1950). Verification of forecasts expressed in terms of probability. Monthly Weather Review, 78, 1-3. https://doi.org/10.1175/1520-0493(1950)078<0001:VOFEIT>2.0.CO;2.

Bussmann, N., Giudici, P., Marinelli, D., & Papenbrock, J. (2021). Explainable Machine Learning in Credit Risk Management. Comput Econ, 57, 203-216. https://doi.org/10.1007/s10614-020-10042-0.

Davis, J., & Goadrich, M. (2006). The relationship between Precision-Recall and ROC curves. In Proceedings of the 23rd International Conference on Machine Learning (pp. 233-240). ACM Press. https://doi.org/10.1145/1143844.1143874.

DeLong, E. R., DeLong, D. M., & Clarke-Pearson, D. L. (1988). Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics, 44, 837-845.

Friedman, J. H. (2001). Greedy function approximation: A gradient boosting machine. The Annals of Statistics, 29, 1189-1232. https://doi.org/10.1214/aos/1013203451.

Fuster, A., Goldsmith-Pinkham, P., Ramadorai, T., & Walther, A. (2022). Predictably Unequal? The Effects of Machine Learning on Credit Markets. The Journal of Finance, 77, 5-47.

Guo, C., Pleiss, G., Sun, Y., & Weinberger, K. Q. (2017). On Calibration of Modern Neural Networks. https://doi.org/10.48550/arXiv.1706.04599.

Hand, D. J., & Henley, W. E. (1997). Statistical Classification Methods in Consumer Credit Scoring: A Review. J R Stat Soc Ser A Stat Soc, 160, 523-541. https://doi.org/10.1111/j.1467-985X.1997.00078.x.

Hardt, M., Price, E., & Srebro, N. (2016). Equality of Opportunity in Supervised Learning. https://doi.org/10.48550/arXiv.1610.02413.

Hollmann, N., Müller, S., Purucker, L., Krishnakumar, A., Körfer, M., Hoo, S. B., Schirrmeister, R. T., & Hutter, F. (2025). Accurate predictions on small data with a tabular foundation model. Nature, 637, 319-326. https://doi.org/10.1038/s41586-024-08328-6.

Hulok, M. (2025). The EU model of AI governance: regulating artificial intelligence through law and policy. ERA Forum, 26, 527-547. https://doi.org/10.1007/s12027-025-00869-1.

Ke, G., Meng, Q., Finley, T., Wang, T., Chen, W., Ma, W., Ye, Q., & Liu, T.-Y. (2017). LightGBM: A Highly Efficient Gradient Boosting Decision Tree. In Advances in Neural Information Processing Systems. Curran Associates, Inc.

Lessmann, S., Baesens, B., Seow, H.-V., Thomas, L. C. (2015). Benchmarking state-of-the-art classification algorithms for credit scoring: An update of research. European Journal of Operational Research, 247, 124-136. https://doi.org/10.1016/j.ejor.2015.05.030.

Lundberg, S., & Lee, S.-I. (2017). A Unified Approach to Interpreting Model Predictions. https://doi.org/10.48550/arXiv.1705.07874.

Molnar, C. (2025). Interpretable Machine Learning: A Guide For Making Black Box Models Explainable. Christoph Molnar, Munich, Germany.

Papagiannidis, E., Mikalef, P., & Conboy, K. (2025). Responsible artificial intelligence governance: A review and research framework. The Journal of Strategic Information Systems, 34, 101885. https://doi.org/10.1016/j.jsis.2024.101885.

Platt, J. (2000). Probabilistic Outputs for Support Vector Machines and Comparisons to Regularized Likelihood Methods. Adv. Large Margin Classif., 10.

Prokhorenkova, L., Gusev, G., Vorobev, A., Dorogush, A.V., & Gulin, A. (2019). CatBoost: unbiased boosting with categorical features. https://doi.org/10.48550/arXiv.1706.09516.

Ridzuan, N. N., Masri, M., Anshari, M., Fitriyani, N. L., & Syafrudin, M. (2024). AI in the Financial Sector: The Line between Innovation, Regulation and Ethical Responsibility. Information, 15, 432. https://doi.org/10.3390/info15080432.

Rudin, C. (2019). Stop Explaining Black Box Machine Learning Models for High Stakes Decisions and Use Interpretable Models Instead. https://doi.org/10.48550/arXiv.1811.10154.

Somvanshi, S., Das, S., Javed, S. A., Antariksa, G., & Hossain, A. (2024). A Survey on Deep Tabular Learning. arXiv.org. https://arxiv.org/abs/2410.12034v1.

Ye, S., Jiang, Y., Chen, B., & Jiang, C. (2025). Credit Behavior Scorecards Based on TabTransformer and Model Interpretability Exploring. In Proceedings of the 2025 4th International Conference on Big Data, Information and Computer Network (pp. 420-428). Association for Computing Machinery. https://doi.org/10.1145/3727353.3727422.

Yeh, I.-C. (2009). Default of credit card clients [Data set]. UCI Machine Learning Repository. https://archive.ics.uci.edu/dataset/350/default+of+credit+card+clients.

Yeh, I.-C., & Lien, C. (2009). The comparisons of data mining techniques for the predictive accuracy of probability of default of credit card clients. Expert Systems with Applications, 36, 2473-2480. https://doi.org/10.1016/j.eswa.2007.12.020.

Yurdakul, B., & Naranjo, J. (2020). Statistical properties of the population stability index. JRMV. https://doi.org/10.21314/JRMV.2020.227.

Zadrozny, B., & Elkan, C. (2002). Transforming classifier scores into accurate multiclass probability estimates. In Proceedings of the Eighth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (pp. 694-699). Association for Computing Machinery. https://doi.org/10.1145/775047.775151.