Sources, licences and method
Where each file comes from, what you may do with it, and how we check the numbers.
Real datasets
| Dataset | Source | Licence | Retrieved |
|---|---|---|---|
| Earthquake data (USGS) | U.S. Geological Survey, Earthquake Hazards Program, real-time feed 2.5_month.csv. Link | U.S. public domain (USGS-authored data) | 2026-11-03 |
| U.S. state population data | U.S. Census Bureau, Population Estimates Program, Vintage 2024 (NST-EST2024-ALLDATA). Link | U.S. public domain (federal government work) | 2026-11-03 |
| World population by country | World Bank, World Development Indicators, SP.POP.TOTL (Population, total). Link | CC BY 4.0 (World Bank open data) | 2026-11-03 |
| GDP by country | World Bank, World Development Indicators, NY.GDP.MKTP.CD (GDP, current US$). Link | CC BY 4.0 (World Bank open data) | 2026-11-03 |
| Daily weather data, 2024 | Open-Meteo Historical Weather API (ERA5-based reanalysis). Link | CC BY 4.0. Weather data by Open-Meteo.com | 2026-11-03 |
What we changed: we kept the columns listed in each data dictionary, renamed them to lower-case names, removed World Bank regional and income aggregates, removed Puerto Rico and summary rows from the Census table, and added the 2023-2024 change columns, which are computed from the Census figures. We did not edit any values. When you use the World Bank or Open-Meteo files, credit the source as their CC BY 4.0 licence asks: for example "Data: World Bank, CC BY 4.0, via DataBrighten" or "Weather data by Open-Meteo.com".
Synthetic datasets
The 13 business tables are generated by our own code. Every value is made up, following fixed rules so the data behaves like the real thing: sales = quantity × unit price × (1 − discount); profit = sales − cost; letter grades follow the average score; invoice status follows the due and paid dates; a bank balance is the opening balance plus every amount. Names are random pairs of common U.S. first names and surnames, emails use the reserved example.com domain, and phone numbers use 555-0100 to 555-0199, which are set aside for fiction. Cities and states are real places; the businesses and people are not.
We release the synthetic files under CC0: use them for anything, no credit needed.
How the answers are checked
Each practice answer is computed twice from the files you download: once with SQL in SQLite, loading our own .sql file, and once with plain Python reading our .csv file. The build stops if the two results differ by more than a rounding cent. The pandas snippets are run on the same files and their real output is shown under each one. Excel formulas use the column letters of the downloaded file.
Corrections
Spotted a wrong number or a broken file? Tell us on the contact page and we will fix it and note the date here.