Date of Award
Spring 5-29-2024
Document Type
Thesis (Undergraduate)
Department
Computer Science
First Advisor
Daniel Rockmore
Abstract
Banks and fintech lenders increasingly rely on computer-aided models in lending decisions. Traditional models were interpretable: decisions were based on observable factors, such as whether a borrower's credit score was above a threshold value, and explainable in terms of combinations of these factors. In contrast, modern machine learning models are opaque and non-interpretable. Their opaqueness and reliance on historical data that is the artifact of past racial discrimination means these new models risk embedding and exacerbating such discrimination, even if lenders do not intend to discriminate. We calibrate two random forest classifiers using publicly available HMDA loan data and publicly available Fannie Mae loan performance data. We use two Explainable Artificial Intelligence (XAI) models, LIME and SHAP, to characterize what features drive the decisions produced by these calibrated ML lending models. Our preliminary findings suggest a significant impact of various racial factors within a model's decision-making process when it has access to such information, as seen in the model trained on HMDA data. These results highlight the need for further investigation to understand and address these influences in depth.
Recommended Citation
Kalikman, Will, "Investigating Bias in Mortgage-Rate Machine Learning Models" (2024). Computer Science Senior Theses. 29.
https://digitalcommons.dartmouth.edu/cs_senior_theses/29