Research Symposium

BIO

I am Sophia Mena and I am majoring in Applied Math with a minor in Spanish. I have been able to do my research through the directed reading program here at Florida State University.

Bias-Aware Machine Learning for Equitable Credit Lending Decisions

Abstract

Biases in machine learning models have significant implications across sectors such as hiring, medical insurance, and auto loans, often negatively affecting marginalized communities the most. This research focuses on developing a model to promote equal credit lending opportunities by identifying applicants with strong credit quality while addressing inherent biases. Using a training and testing dataset, we employ a logistic regression model trained on both quantitative features (e.g., FICO score, payment-to-income ratio, loan-to-value ratio, etc.) and qualitative features (e.g., gender, race). One of primary objectives is to identify and mitigate biases within the model, particularly those affecting marginalized groups such as Black and Hispanic applicants. This research also aims to accurately qualify candidates with a good credit quality. From preliminary analysis, we anticipate for features like FICO score, total number of never delinquent or derogatory trades, and never delinquent trades reported significantly influence approval odds, while biases are expected to emerge among features like race and gender. These findings can provide a framework for financial institutions to improve their credit approval systems, promoting equitable access among all applicants.

Keywords: Machine Learning and Bias