Comparative Analysis of Generalization for Machine Learning Application in Loan Default Binary Classification
Jiayuan Ma
2024
Abstract
Predicting loan defaults is critical for banks because it enables financial institutions to assess the risks associated with loan approval. While traditional methods of evaluating loan applicants rely on subjective assessments, Machine Learning (ML) has emerged as a powerful alternative to predicting defaults. This study conducted a comparative analysis of six ML models on a dataset of 255,347 loan applicants. The goal is to evaluate the generalization ability of each model when exposed to different data distributions. The results show significant performance degradation across all models when tested on unseen data, highlighting the issue of distribution shifts between training and testing sets. Some techniques such as domain adaptation and distribution alignment are discussed as potential solutions to improve model robustness. These findings provide valuable insights that could guide financial institutions in selecting more reliable, adaptable, and robust models for accurately predicting loan defaults, thus improving decision-making processes and reducing financial risk.
DownloadPaper Citation
in Harvard Style
Ma J. (2024). Comparative Analysis of Generalization for Machine Learning Application in Loan Default Binary Classification. In Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI; ISBN 978-989-758-726-9, SciTePress, pages 319-323. DOI: 10.5220/0013224800004568
in Bibtex Style
@conference{ecai24,
author={Jiayuan Ma},
title={Comparative Analysis of Generalization for Machine Learning Application in Loan Default Binary Classification},
booktitle={Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI},
year={2024},
pages={319-323},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013224800004568},
isbn={978-989-758-726-9},
}
in EndNote Style
TY - CONF
JO - Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI
TI - Comparative Analysis of Generalization for Machine Learning Application in Loan Default Binary Classification
SN - 978-989-758-726-9
AU - Ma J.
PY - 2024
SP - 319
EP - 323
DO - 10.5220/0013224800004568
PB - SciTePress