Open Access
Subscription Access
Enhancing Loan Default Prediction in Banking Using Light Gradient Boosting Model and Explainable AI
Abstract
Banks offer loan to a person after using a credit scoring model which is best when it is explained. In this paper we have created a similar Light Gradient Boosting model for predicting credit default on a unique dataset which we have been collected from Credit Bureau Data Inc. Here, not only we have used the LightGBM model but we also combined it with SHAP that falls under the Explainable AI concept, which helps us to identify the most important features based, on which the model has predicted an outcome. The LightGBM model is clearly better than the normal Logistic Regression which is generally used by the banks nowadays. The LightGBM model predicts the most important features are `person income`, `loan interest rate`, `loan amount`, `loan present income`, `person employment length`, `person age`, etc. Our main motive here is to implement the usage of Explainable AI in the banking and insurance sectors and improve the interpretability and reliability of the banks`
models.
Keywords
Disclaimer/Regarding indexing issue:
We have provided the online access of all issues and papers to the indexing agencies (as given on journal web site). It’s depend on indexing agencies when, how and what manner they can index or not. Hence, we like to inform that on the basis of earlier indexing, we can’t predict the today or future indexing policy of third party (i.e. indexing agencies) as they have right to discontinue any journal at any time without prior information to the journal. So, please neither sends any question nor expects any answer from us on the behalf of third party i.e. indexing agencies.Hence, we will not issue any certificate or letter for indexing issue. Our role is just to provide the online access to them. So we do properly this and one can visit indexing agencies website to get the authentic information. Also: DOI is paid service which provided by a third party. Journal never mentioned that we have DOI number. However, to get free DOI, author can register your work which published with Zonodo (https://zenodo.org/signup/). We have no objection for this open access repository.