Comparative Analysis of Bank Loan Risk Forecasting Based on Machine Learning Models Versus SHAP Interpretation

DOI

10.2991/978-94-6239-801-6_157How to use a DOI?

Keywords

Loan Default Prediction; Machine Learning; Ensemble Learning; SHAP Interpretation; Model Robustness

Abstract

With the rapid development of the social economy, loan scales have expanded quickly. While banks and other financial institutions can generate considerable profits, this also means a significant increase in the risk of loan defaults. Given the vast amount of high-dimensional customer data, machine learning technology needs to be introduced. Based on 148,670 loan records from the Kaggle platform, this study systematically compared the performance of eight models—logistic regression, decision trees, random forests, gradient boosting, XGBoost, LightGBM, and soft voting and stacked integration—in default prediction, and model stability is evaluated through 5-fold cross-validation, the Shapley Additive Explanations method (SHAP) is used to interpret global feature importance, and robustness is verified from two dimensions: grouping tests based on loan amount and noise resistance tests on standardized interest rates. The results show that tree-based ensemble models are significantly superior to logistic regression models; Random Forest achieved the highest Area Under Curve (AUC) but low recall, while stacked integration achieved a better balance between accuracy and recall; Interest rates, credit types, negative amortization, debt income, and loan-to-valueratios are key risk characteristics; Interest ratecharacteristics are highly sensitive to noise, while the model performs stably across different loan amount groups.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Yuning Wang
PY  - 2026
DA  - 2026/10/23
TI  - Comparative Analysis of Bank Loan Risk Forecasting Based on Machine Learning Models Versus SHAP Interpretation
BT  - Proceedings of the 2026 2nd International Conference on Financial Innovation and Marketing Management (FIMM 2026)
PB  - Atlantis Press
SP  - 1492
EP  - 1501
SN  - 2352-5428
UR  - https://doi.org/10.2991/978-94-6239-801-6_157
DO  - 10.2991/978-94-6239-801-6_157
ID  - Wang2026
ER  -