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.2026.1.1.004

Ngày nhận bài: 12/02/2026

Ngày nhận bài sửa: 02/04/2026

Ngày duyệt đăng: 12/04/2026

Email: minhhb@hub.edu.vn

Đơn vị công tác:

Lượt xem: 158

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

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.