Data, based in reality.
Datasets I collected, cleaned or rebuilt, each with a datasheet: what it measures, where it came from, its limits, an example query, and a download. Read it, query it in the browser, or print it.
Raw captures never leave my machine. A manifest names the columns a release may contain; a script builds the public files from it; the checks run in public. How that works.
GB electricity carbon intensity
Half-hourly carbon intensity of electricity for Great Britain and its seventeen distribution regions, with the forecast generation mix. Republished from NESO’s Carbon Intensity API as flat files you can query without an API key.
South Wales grid, by the half hour
Region 7 cut of DS-001 with the generation mix pivoted to columns, for the notebook’s local questions.
Household heating, two dwellings
Flow, return and room temperatures with outdoor weather, and boiler state where it was recorded, over a heating year.
Supplier half-hourly consumption
Gas and electricity import from the supplier’s smart-meter feed, aggregated to daily so occupancy isn’t readable.
One solar panel, metered
Voltage, current and power from an INA228 monitor on a small panel, with plausibility flags.
Food price inflation, by country
World Bank real-time food price monitoring, cleaned and joined for a coursework analysis.
Porthcawl surf conditions
Swell, wind and tide for one beach, for a question about when the forecast is worth trusting.
Ask, and say what you’d do with it. Requests decide the order.
Email JulianThe notebook these datasets feed is at julian-elliott.github.io. Source for this site: GitHub.