From Credit Scoring to Artificial Intelligence-Driven Loan Default Prediction

DOI

10.2991/978-94-6239-648-7_45How to use a DOI?

Keywords

Loan Default Prediction; Credit Scoring; Machine Learning; Deep Learning

Abstract

Loan default prediction is important to both the credit allocation and portfolio risk management. The conventional scorecard is based on predefined handcrafted features and linear assumptions, thus ignoring nonlinear and temporal characteristics of borrowers behaviors. Article introduce a novel sandwiched framework along with two interesting instantiations. This article consider these standard interpretable feature based predictors as strong baselines. To model richer structure, this article devise three neural modules: (i) a Deep Neural Network (DNN) to model nonlinear interactions among high-dimensional attributes; (ii) a Long Short-Term Memory (LSTM) network to model repayment sequences in order to capture temporal patterns; and (iii) a Graph Neural Network (GNN) to model borrower–merchant or borrower–borrower relations in order to capture network dependences. Model interpretability is evaluated through SHAP to bring the predictions back to important financial features and to validate with domain experience. On both two different data sets, the hybrid methods achieve better discrimination and default recall than the classical benchmarks, in particular, LSTM and GNN are strong for temporal and relational signals. SHAP identifies cash flow homogeneity and income stability as the main contributors. The framework strikes a good balance between accuracy and interpretability which makes it appropriate for risk management in digital lending.

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  - Zaiyu Zhang
PY  - 2026
DA  - 2026/04/24
TI  - From Credit Scoring to Artificial Intelligence-Driven Loan Default Prediction
BT  - Proceedings of the International Workshop on Advances in Deep Learning for Image Analysis and Computer Vision (IWADIC 2025)
PB  - Atlantis Press
SP  - 409
EP  - 417
SN  - 2352-538X
UR  - https://doi.org/10.2991/978-94-6239-648-7_45
DO  - 10.2991/978-94-6239-648-7_45
ID  - Zhang2026
ER  -