How the data was put together
This page summarizes where the numbers and pages come from, how they flow through the fuelprices project, and who did what. It is meant for readers who want context on methodology and attribution—not as a legal or official data warranty.
Fuel prices
Daily state-average retail gasoline prices (regular and other grades) are scraped from the American Automobile Association (AAA) public table and stored as Parquet files by state. Weekly retail gasoline for a limited set of U.S. areas comes from the U.S. Energy Information Administration (EIA) Open Data API, also saved as Parquet. R and Python scripts merge, deduplicate by date, and render charts and HTML reports. Exact URLs and series IDs appear on the individual report pages and in the project source.
State government & elections
The state government atlas (including Electoral College tables, trifecta labels, and 2026 seat summaries) is produced from JSON inputs and Python build scripts in the repository. Population and apportionment figures trace to Census 2020 and published House apportionment; chamber balances and election notes are snapshots that should be verified against official state sources as elections unfold.
This hub (election_fuel_web)
The hub links static HTML outputs (fuel reports + atlas pages). A sync script can copy those files into this folder so the site can be previewed or deployed as a self-contained bundle. Navigation and explanatory copy are tuned for the election-context narrative (fuel as a pocketbook issue alongside divided-government states).
Attribution
Crafted by ML. Project design, data pipeline, scrapers and fetchers, Parquet layout, R reporting, atlas data and build scripts, and the overall election-fuel narrative are ML’s work.
Data fetched, presented, and worded with Cursor AI. Cursor AI assisted with implementation details, HTML/CSS for these pages, report wording, refactoring, and tooling (including generators and the hub); ML reviewed and owns the substantive choices.