A novel deep learning approach to enhance creditworthiness evaluation and ethical lending practices in the economy

Xiaoyan Qian, Helen Huifen Cai, Nisreen Innab, Danni Wang (), Tiziana Ciano and Ali Ahmadian
Additional contact information
Xiaoyan Qian: Nanjing University
Helen Huifen Cai: Middlesex University Business School
Nisreen Innab: AlMaarefa University
Danni Wang: Wenzhou Business College
Tiziana Ciano: University of Aosta Valley
Ali Ahmadian: Mediterranea University of Reggio Calabria

Annals of Operations Research, 2025, vol. 346, issue 2, No 32, 1597-1619

Abstract: Abstract Evaluating a borrower's creditworthiness and enabling ethical lending practices are two of the most essential functions of credit scoring, making it an integral part of the economy. Credit risk management is an essential aspect of the financial industry, with the primary goal of minimising potential losses caused by customers failing to meet their credit responsibilities, such as fails to pay and bankruptcies. This risk is inherent in lending activities, where lenders extend credit to individuals or businesses. The traditional credit scoring approaches, which rely on statistical and machine learning techniques to analyse complex data and non-linear correlations in credit data has to be improved. Because the current financial sector lacks credit scoring, a deep learning network-based credit ranking model is presented in this research. This paper applies the complicated field of deep learning known as the stacked unidirectional and bidirectional long short-term memory model in the network to resolve credit scoring issues. Since scoring is not a time sequence issue, the suggested model uses the three-layer stacked LSTM and bidirectional LSTM architecture by modelling public datasets in a new way. Our suggested models beat state-of-the-art, considerably more difficult deep learning methods, proving that we could keep complexity to a minimum. The research findings indicate that the model demonstrates high levels of accuracy across various datasets. The model obtains an accuracy of 99.5% on the Australian dataset, 99.4% on the German dataset (categorical), 99.7% on the German dataset (numerical), 99.2% on the Japanese dataset, and 99.8% on the Taiwanese dataset. These results highlight the robustness and effectiveness of the model in accurately predicting outcomes for different geographical regions.

Keywords: Credit scoring; Deep learning; Long short-term memory (LSTM); Economy; Credit risk management; Financial (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://link.springer.com/10.1007/s10479-024-05849-1 Abstract (text/html)
Access to the full text of the articles in this series is restricted.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:annopr:v:346:y:2025:i:2:d:10.1007_s10479-024-05849-1

Ordering information: This journal article can be ordered from
http://www.springer.com/journal/10479

DOI: 10.1007/s10479-024-05849-1

Access Statistics for this article

Annals of Operations Research is currently edited by Endre Boros

More articles in Annals of Operations Research from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().