Author:
M N. Sowmiya,S. Jaya Sri,S. Deepshika,G. Hanushya Devi
Abstract
The proposed research focuses on enhancing the interpretability of risk evaluation in credit approvals within the banking sector. This work employs LIME (Local Interpretable Model-agnostic Explanations) and SHAP (SHapley Additive exPlanations) to provide explanations for individual predictions: LIME approximates the model locally with an interpretable model, while SHAP offers insights into the contribution of each feature to the prediction through both global and local explanations. The research integrates gradient boosting algorithms (XGBoost, LightGBM) and Random Forest with these Explainable Artificial Intelligence (XAI) techniques to present a more comprehensible framework. The results demonstrate how interpretability methods such as LIME and SHAP enhance the transparency and trustworthiness of machine learning models, which is crucial for applications in credit risk evaluation.
Publisher
Inventive Research Organization
Reference16 articles.
1. [1] Pandey, Trilok Nath, Alok Kumar Jagadev, Suman Kumar Mohapatra, and Satchidananda Dehuri. "Credit risk analysis using machine learning classifiers." In 2017 International Conference on Energy, Communication, Data Analytics and Soft Computing (ICECDS), Chennai, India pp. 1850-1854. IEEE, 2017.
2. [2] Misheva, Branka Hadji, Joerg Osterrieder, Ali Hirsa, Onkar Kulkarni, and Stephen Fung Lin. "Explainable AI in credit risk management." arXiv preprint arXiv:2103.00949 (2021).
3. [3] Bussmann, Niklas, Paolo Giudici, Dimitri Marinelli, and Jochen Papenbrock. "Explainable machine learning in credit risk management." Computational Economics 57, no. 1 (2021): 203-216.
4. [4] Bussmann, N., Giudici, P., Marinelli, D., & Papen brock, J. Explainable AI in fintech risk management. Frontiers in Artificial Intelligence, April 2020, vol 3, 26.
5. [5] Biecek, Przemysław, Marcin Chlebus, Janusz Gajda, Alicja Gosiewska, Anna Kozak, Dominik Ogonowski, Jakub Sztachelski, and Piotr Wojewnik. "Enabling machine learning algorithms for credit scoring--explainable artificial intelligence (XAI) methods for clear understanding complex predictive models." arXiv preprint arXiv:2104.06735 (2021).