ipeds

A Python package for working with IPEDS data

Ask natural-language questions about U.S. colleges (IPEDS) and get answers with tables and charts — a self-hosted FastAPI + React app with an LLM NL→SQL agent.

48,073 of the 169,578 US college degree and certificate programs conferring in 2013 had stopped by 2021, while the total count grew. Pipeline, checks, and the published page.

Data analysis of U.S. university budget allocation patterns (2018-2023). Analyzing why college tuition is expensive and where the money goes.

Beta, pre-release. Grades US higher-education institutions on what they disclose, not on how they perform. Every published value is classified as reported, implausible, suppressed, not applicable, or missing, and per-field disclosure drift between IPEDS and College Scorecard snapshots is measured in both directions.

Reproducible NCES/IPEDS higher-education research brief with QA, state profiles, methodology crosswalk, and recurring-report operations.

download NCES IPEDS data

Python + DuckDB pipeline for downloading, harmonizing, QC-validating, and panelizing NCES IPEDS data into research-ready institution-year datasets.

A proof-of-concept for using RAG to search the IPEDS metatdata.

Build a harmonized DuckDB of 27 million IPEDS records from 1997–2024 for easy SQL queries on U.S. higher-education data.