Installation

Use Python 3.11 or newer. You do not need R or rpy2 to fit models, predict or run inference.

Install a release wheel

Download the wheel for your platform from the latest release. The published wheel has been checked on macOS ARM64, including Apple Silicon Macs. In the directory containing the downloaded wheel, run:

python -m venv .venv
source .venv/bin/activate
python -m pip install ./rparity-*.whl

Select a compatible wheel rather than renaming a file for another platform. If there is no wheel for your platform, use a source installation and check your intended analysis locally. Execution on other platforms has not yet been validated.

Install from source

git clone https://github.com/jbaehova/rparity.git
cd rparity
python -m pip install .

This installs the library into your current Python environment. To isolate it, create and activate a virtual environment first. Package builds obtain the native OpenBLAS backend used for cubic shrinkage smooths; its libraries and license notices are included in wheels.

Check your installation

python -c "from rparity import lmer, gam, glmmTMB, emmeans; print('rparity is ready')"

The five declared Python dependencies are NumPy, SciPy, pandas, formulaic and statsmodels. pandas data frames work throughout the public API. Polars input is also accepted and converted to pandas internally; install Polars separately if you use it.

Continue with the quickstart, or run an analysis from the repository's examples directory.