13 - Predict the production of distributed PV-assets
🌞 Your task is to predict the total solar production of thousands of hidden PV systems across Switzerland. Armed only with weather data from MeteoSwiss and past feed-in values, you must uncover the patterns behind the panels — despite knowing almost nothing about the actual PV sites themselves.
Challenge Owner
Postdate 01.09.2025
Switzerland’s solar capacity has been growing at record speed, with thousands of new PV installations added every year. This surge in distributed generation strengthens the renewable share in the energy mix — but it also makes forecasting total PV feed-in a real challenge.Â
The difficulty: We don’t know the exact locations, sizes, or orientations of the many small systems.Â
Your mission: Develop a model to predict PV feed-in for January to August 2025 at 15-minute resolution. You’ll have access to historical feed-in data from 2020 to 2024, as well as detailed weather measurements from MeteoSwiss.Â
Use the past to train your model, then test it by forecasting 2025. At the end, your predictions will be compared to the real feed-in data for 2025. The team with the most accurate model wins — can you crack the code of Switzerland’s solar future? ⚡🌞