HMDA

The HMDA dataset, published by the Consumer Financial Protection Bureau (CFPB), contains detailed information on individual mortgage applications submitted to covered financial institutions in the United States. The data includes applicant demographics (e.g., race, sex, and income), loan characteristics (e.g., amount, purpose, interest rate), and underwriting outcomes (e.g., loan approval or denial). The dataset supports regulatory compliance, transparency in lending, and research on fair housing and access to credit.The HMDA Modified Loan\/Application Register (LAR) anonymizes personally identifiable information while retaining the core attributes necessary for analyzing patterns of access, equity, and potential discrimination in mortgage lending. It is widely used in academic and policy research to study disparities across racial, gender, and income groups, and has become a benchmark dataset for evaluating fairness-aware machine learning methods in financial decision-making.This dataset forms part of the empirical foundation for studies focused on bias detection, fairness metrics (e.g., disparate impact, statistical parity), and algorithmic transparency in predictive modeling for loan approval. The 2022 release includes over 20 million records submitted by thousands of reporting institutions and remains one of the most comprehensive public datasets on lending activity in the U.S. housing market.