Installation¶

From PyPI (recommended)¶

pip install pinball-qr

From Source¶

Pinball includes compiled Fortran extensions, so building from source requires a Fortran compiler (e.g. gfortran) and the Meson build system.

git clone https://github.com/mhoward2718/pinball-qr.git
cd pinball-qr
python -m venv .venv
source .venv/bin/activate
pip install meson-python meson ninja numpy
pip install -e . --no-build-isolation

macOS¶

brew install gcc  # provides gfortran

Ubuntu / Debian¶

sudo apt install gfortran

Dependencies¶

Package	Version	Purpose
NumPy	≥ 1.23	Array operations
SciPy	≥ 1.10	Statistical distributions, sparse linear algebra
scikit-learn	≥ 1.3	Base estimator classes, validation utilities

Verify Installation¶

import pinball
print(pinball.__version__)

# Quick smoke test — fit the Engel dataset
from pinball import QuantileRegressor, load_engel
X, y = load_engel(return_X_y=True)
model = QuantileRegressor(tau=0.5).fit(X, y)
print(f"Intercept: {model.intercept_:.2f}, Slope: {model.coef_[0]:.4f}")