DemandCast

Global hourly electricity demand forecasting

Supported by

About

DemandCast is a Python-based project focused on collecting, processing, and forecasting hourly electricity demand data. The aim of this project is to support energy planning studies by using machine learning models to generate hourly time series of future electricity demand or for countries without available data.

Features

• Retrieval of open hourly and sub-hourly electricity demand data from public sources (ETL).
• Retrieval of weather and socio-economic data (ETL).
• Forecasting using machine learning models (models).
• Modular design for adding new countries or data sources.
• Support for reproducible, containerized development.

The project is in active development, we are always looking for suggestions and contributions!

Repository structure

demandcast/
├── docs/                   # Project documentation (MkDocs)
├── ETL/                    # Scripts for extracting, transforming, and loading data
├── models/                 # Machine learning models for demand forecasting
├── .devcontainer/          # Development container configuration
├── .github/                # Github specifics such as actions
├── .gitignore              # File lists that git ignores
├── .pre-commit-config.yaml # Pre-commit configuration
├── .python-version         # Python version
├── CONTRIBUTING.md         # Guide to contributing
├── Dockerfile              # Docker setup for containerized runs
├── mkdocs.yml              # Documentation configuration file
├── pyproject.toml          # Project metadata and dependencies
├── ruff.toml               # Ruff configuration
└── uv.lock                 # Lockfile for project's dependencies

Data collection progress

Find the code that we used to retrieve the data in their respective files inside the ETL folder.

Getting started

1. Clone the repository

git clone https://github.com/open-energy-transition/demandcast.git
cd demandcast

2. Set up your environment

This project uses uv as a pacakge manager. It can be used within the provided Dockerfile or installed standalone (see installing uv)

uv sync

Alternatively, you may use conda to install the packages listed in pyproject.toml.

3. Run scripts

Scripts can be run directly using:

uv run path/to/script.py

If you have chosen conda as your package manager, you can run scripts with:

python path/to/script.py

Jupyter notebooks (details) can be launched with:

uv run --with jupyter jupyter lab --allow-root

Development workflow

Run tests and check test coverage

uv run pytest
uv run pytest --cov --cov-report term-missing

Pre-commit and lint code

uvx ruff format
uvx ruff check --fix
uvx mypy
uvx pre-commit

Documentation

The documentation is currently hosted on GitHub pages. It is built with mkdocs.

To run it locally:

uv run mkdocs serve

Contributing

We welcome contributions in the form of:

• Country-specific ETL modules
• New or improved forecasting models
• Documentation and testing enhancements

Please follow the repository’s structure and submit your changes via pull request.

License

This project is licensed under the GNU Affero General Public License v3.0 (AGPL-3.0).