The repository contains a solution to predict the energy behavior of prosumers in order to minimize the energy imbalance. Prosumers are individuals who both consume and produce energy.
The energy imbalance is a state where the amounts of energy production and consumption differ. The negative effects of energy imbalace are increased operational costs, potential energy grid instability, and inefficient use of energy resources. By predicting the energy behavior of prosumers the energy imbalance could be decreased, thus lowering the operational costs and stabilizing the grid system.
The problem was initially presented by Eesti Energia on Kaggle competition.
- Clone this repository using
git clone https://github.com/bonskotti/predict-energy-behavior.git - Install requirements using
pip install -r requirements.txt - Run
jupyter-lab predict-energy-behavior-of-prosumers-predictions.ipynb
- prosumer.csv - Prosumer data
- gas_prices.csv - Gas price data
- client.csv - Energy company client data
- electricity_prices.csv - Electricity price data
- weather_forecast.csv - Weather forecast data
- weather_history.csv - Weather history data
- weather_station_to_county_mapping.csv - Weather station data
- county_id_to_name_map.json - County data
All data was provided by Eesti Energia.
NOTE: This repository contains only a small subset of the original data. To access the full dataset, visit the Kaggle competition website.
Predicted energy amounts for a sample subset of data can be seen in the figure above. The predicted amounts follow the actual amounts ("Ground truth") relatively well.