Skip to content

Installation

Core install

pip install xdflow

This installs the core runtime for metadata-driven pipelines built on xarray, numpy, pandas, scikit-learn, scipy, and joblib.

Optional extras

Install only the dependency sets you need:

pip install "xdflow[tuning]"
pip install "xdflow[lightgbm]"
pip install "xdflow[mlflow]"
pip install "xdflow[viz]"
pip install "xdflow[spectral]"
pip install "xdflow[adaptation]"
pip install "xdflow[all]"

Current extras in pyproject.toml:

  • tuning: Optuna-backed tuning support. Tuning is a core XDFlow workflow; the extra only keeps optimizer dependencies out of the minimal install.
  • lightgbm: LightGBM predictor wrapper
  • mlflow: experiment tracking integration
  • viz: plotting helpers
  • spectral: spectral-analysis dependencies
  • adaptation: domain-adaptation dependencies
  • all: all optional runtime extras

Development install

For editable development with linting and test tools:

python -m pip install -e ".[dev]"

This repository also includes a uv.lock, so a reproducible local setup can be created with:

uv sync --extra dev --extra docs

For tuning development or docs examples:

uv sync --extra dev --extra docs --extra tuning

Run the core quickstart example from the repository root:

uv run python examples/quickstart.py

To run the spectral tutorial examples locally, include the spectral extra:

uv sync --extra dev --extra docs --extra spectral

Documentation build

Local docs build:

uv run mkdocs build --strict

The published docs are configured for Read the Docs through the top-level .readthedocs.yaml and mkdocs.yml.