hmda

Build a harmonized DuckDB database from 40+ years of HMDA mortgage data (685M rows) in a single command.

HMDA mortgage lending disparity analyzer — denial rates, racial disparities, lending deserts, and lender benchmarking

Independent model validation and AI governance framework for high-stakes financial decision models using public HMDA mortgage data.

Statistical lending-disparity screening for public HMDA data — a transparent, open-source screening tool for community advocates and journalists. Not an examination methodology

AI skills for NMTC eligibility, bank-CDFI peer benchmarking & HMDA analysis — grounded in audited PyPI tools, not hallucinated.

Predicts U.S. home loan approvals using 4.25M real HMDA 2023 applications — XGBoost, Random Forest, Logistic Regression, ROC-AUC 0.9932

Orchestrated HMDA mortgage data pipeline using Apache Airflow — automated ingestion, cleaning, quality validation, and weekly model performance monitoring on 279K NC loan applications.

A simple, interpretable credit approval model built on HMDA data using decision trees

Final Project for INFO 254 at UC Berkeley in Spring 2018

End-to-end mortgage approval analytics using Python, DuckDB, SQL, Tableau, and 3.8 million HMDA records.

Analysis code for HMDA loan-level mortgage data and FFIEC Census files: lending disparities, tract-boundary corrections across vintages, longitudinal series, systemic-bank metrics, and a Python replication of published R methodology. Code-only; no data shipped.

Analyzing home mortgage data.

Mortgage denial-rate analysis (HMDA 2024) — Python + SQL + Tableau

An end-to-end ELT pipeline using DuckDB, dbt and Streamlit for fair lending analysis.

ELT pipeline and Postgres warehouse for the public HMDA mortgage data. Loads 23.7 million loan records across 2023 and 2024 through bronze, silver and gold layers, with fifteen quality gates.

End-to-end mortgage credit risk analytics using Python, SQL, and Power BI on 2024 HMDA data.