Explainable Machine Learning Models of Consumer Credit Risk

The Journal of Financial Data Science, Fall 2023, 5 (4) 9 - 39 DOI: 10.3905/jfds.2023.1.141

Posted: 14 Jan 2022 Last revised: 25 Apr 2022

See all articles by Randall Davis

Randall Davis

Massachusetts Institute of Technology (MIT)

Andrew W. Lo

Massachusetts Institute of Technology (MIT) - Laboratory for Financial Engineering

Sudhanshu Mishra

affiliation not provided to SSRN

Arash Nourian

affiliation not provided to SSRN

Manish Singh

Massachusetts Institute of Technology (MIT) - Electrical Engineering and Computer Science

Nicholas Wu

Massachusetts Institute of Technology (MIT)

Ruixun Zhang

Peking University; MIT Laboratory for Financial Engineering

Date Written: January 12, 2022

Abstract

In this paper, we create machine learning (ML) models to forecast home equity credit risk for individuals using a real-world dataset and demonstrate methods to explain the output of these ML models to make them more accessible to the end-user. We analyze the explainability of these models for various stakeholders: loan companies, regulators, loan applicants, and data scientists, incorporating their different requirements with respect to explanations. For loan companies, we generate explanations for every model prediction of creditworthiness. For regulators, we perform a stress test for extreme scenarios. For loan applicants, we generate diverse counterfactuals to guide them with steps to reverse the model's classification. Finally, for data scientists, we generate simple rules that accurately explain 70-72% of the dataset. Our work is intended to accelerate the adoption of ML techniques in domains that would benefit from explanations of their predictions.

Keywords: Machine Learning; Interpretability; Explainable AI; Credit Lending; Inductive Logic Programming; Optimal Trees; LIME; SHAP; Counterfactual.

JEL Classification: C55, C45, G0, G2, G51, C69

Suggested Citation

Davis, Randall and Lo, Andrew W. and Mishra, Sudhanshu and Nourian, Arash and Singh, Manish and Wu, Nicholas and Zhang, Ruixun, Explainable Machine Learning Models of Consumer Credit Risk (January 12, 2022). The Journal of Financial Data Science, Fall 2023, 5 (4) 9 - 39 DOI: 10.3905/jfds.2023.1.141, Available at SSRN: https://ssrn.com/abstract=4006840 or http://dx.doi.org/10.2139/ssrn.4006840

Randall Davis

Massachusetts Institute of Technology (MIT) ( email )

77 Massachusetts Avenue
50 Memorial Drive
Cambridge, MA 02139-4307
United States

Andrew W. Lo

Massachusetts Institute of Technology (MIT) - Laboratory for Financial Engineering ( email )

100 Main Street
E62-618
Cambridge, MA 02142
United States
617-253-0920 (Phone)
781 891-9783 (Fax)

HOME PAGE: http://web.mit.edu/alo/www

Sudhanshu Mishra

affiliation not provided to SSRN

Arash Nourian

affiliation not provided to SSRN

Manish Singh (Contact Author)

Massachusetts Institute of Technology (MIT) - Electrical Engineering and Computer Science ( email )

77 Massachusetts Avenue
Cambridge, MA 02139-4307
United States

Nicholas Wu

Massachusetts Institute of Technology (MIT) ( email )

77 Massachusetts Avenue
50 Memorial Drive
Cambridge, MA 02139-4307
United States

Ruixun Zhang

Peking University ( email )

5 Yiheyuan Road
Haidian District
Beijing, Beijing 100871
China

MIT Laboratory for Financial Engineering ( email )

100 Main Street
E62-611
Cambridge, MA 02142

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Abstract Views
4,054
PlumX Metrics