Team "Voltcasters": Christine, Ferdinand, Moritz, Jerome
Forecasting the wind energy production will grow in importance as wind energy is one of the fast-growing renewable energy sources in the world. In this capstone we worked in a group of 4 people. The goal was to predict the wind energy generation 24 h ahead with an hourly resolution, for 10 wind farms located in Australia, based on wind forecasts. We built several machine learning models in Python, to find our best model. This included extensive explanatory data analysis, feature engineering as well as finding an appropriate metric. We developed a dashboard to visualize the results of our final model.
The slides of our presentation: slides
The project has several notebooks:
- EDA (see here)
- The baseline model (see here)
- The models we used (see here)
- The error analysis is done in here and here
We developed a dashboard to visualize our results and deployed it to Heroku:
Energy Output Forecast for the Next 24 Hours
Dashboard Repository:
https://github.com/fklein21/windpower_dashboard
The Data is from the 2014 Global Energy Forecasting Competition.
~ 18k data rows per wind farm
pyenv local 3.9.4
python -m venv .venv
source .venv/bin/activate
pip install --upgrade pip
pip install -r requirements_dev.txt