SHS Web Conf.
Volume 181, 20242023 International Conference on Digital Economy and Business Administration (ICDEBA 2023)
|Number of page(s)
|Financial Analysis and Stock Market Strategies
|17 January 2024
Research on loan approval and credit risk based on the comparison of Machine learning models
University of Melbourne, Melbourne Business School, 200 Leicester Street, Australia
* Corresponding author: chunyy@student.Unimelb.edu.au
Nowadays, home loan is a frequently accessed component of people’s financing activities. Homeowners wants to increase the probability of loan acceptance, however banks seek to borrow money to low risk customers. This paper compared and examined the machine learning models to select when loan applicants evaluating their probability of success. This paper introduced the recommended models for the problem, explanations on how to use the selected model. 6 candidate models, including Logistic regression, Decision tree, Random Forest, support vector machine (SVM), Ada Boost and Neural Network are selected. The model selection process would focus on the model’s accuracy on test data as well as the interpretability of these models. The models’ result was interpreted to derive optimal strategies to be undertaken by both debtors and creditors. Throughout comparison between these models, logistic regression was the best in terms of interpretability and accuracy. Nonetheless, other models could bolster the decision-making process by examining their confusion matrices and the fitted importance of predictors in each model. This paper revealed practical implications of machine learning theories on home loan approval and credit risk and aimed to help decision making for both debtors and creditors.
© The Authors, published by EDP Sciences, 2024
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.