φρόνησιςphronesis — Aristotle's word for practical wisdom: good judgment, and the smart action that follows from it.
Decide well on energy — forecasting and optimisation you can check.
One place for energy data, forecasting and optimisation. Browse 85 data sources and 111 forecasting methods, see them run on real grid and weather history, and upload your own series. Every result records where its data came from and which code produced it.
Four ways to use the platform
Start with what is already here, then bring your own data. Each card opens the thing it describes.
Explore the Atlas
The full catalogue of data sources, forecasting methods, solvers and power networks. Each entry shows how often it updates, what licence it carries, and whether it is ready to use or still being built.
View, plot and export data
Open any series — electricity demand, prices, generation or weather. Chart it over any date range and download the numbers as CSV or JSON.
See forecasting examples
Compare methods on the same problem, tested against real history. You see each one's accuracy, its uncertainty range, and the exact data it was scored on.
Use optimisation examples
Worked decision problems: when to charge a battery, which plants to run, where to build capacity, how power flows through a grid — each solved on a complete example network.
Bring your own data
The catalogue is the starting point. The goal is your own data running through the same tools. Here is how far that works today.
1 · Upload
Upload a CSV, or send readings through the API. Nothing is saved until the file passes its checks.
POST /v1/data/uploads
2 · Check and plot
Missing hours, duplicated timestamps and wrong units are flagged first. Then chart your series next to any catalogue series, and export it.
/validate · /points · /points:export
3 · Forecast your series
Run a working method on one of your own series and get a forecast back, trained on that series alone. Your data never enters a shared benchmark.
POST /v1/forecasts/runs · series_id
4 · Optimise with your data
Put your own assets, prices and limits into the optimisation problems. Partly there: you can size a battery yourself, but prices and limits still come from the market data and the example networks.
Why step 4 says “on the roadmap”. Your uploads are kept separate from the public data on purpose, so no customer's numbers can end up in a shared benchmark. Forecasting keeps that separation by training on your series alone; opening the optimisers to your own data has to keep it too. We would rather tell you what is not ready than let you find out later.
From raw grid data to a decision you can check
One path, four stages. Every stage records what it used, which is what makes the final answer checkable.
Ingest
Collect real measurements from grid operators, markets and weather archives. Every reading is dated and labelled by source. Made-up data is rejected, never quietly mixed in.
Forecast
Test any method — statistical, machine-learning or deep-learning — against real history, moving forward one step at a time. Accuracy is measured on every window, not a flattering one.
Optimise
Turn those forecasts into decisions: when to charge a battery, which plants to run, where to add capacity — solved on complete example power networks.
Decide
Get the answer through an API you can re-run. Every result records the data it used and the exact code version that produced it, so any number can be traced back.
Every entry says how ready it is
Nothing here is counted as working until it is. Each catalogue entry carries one of three labels, and views hide the unfinished ones by default — a toggle shows everything.
Runs end to end
The data source really connects, and the method really runs its own algorithm on real history.
Real, but generic
You get a real forecast, but from a shared fallback model — the method's own algorithm is not running yet.
Registered, pending
Listed so its name and licence stay fixed, but not built yet.
Markets live today
Greece is the main market (ENTSO-E and IPTO, including its ISP results). Every other zone runs through the same pipeline.
Run your first forecast now
A Greek electricity demand forecast is set up and ready to run — about a minute from start to finish. No credit card.