Scikit-Stan is a package of Stan models wrapped in a Scikit-Learn style interface.
This package is currently under active development and should be treated as beta software.
Documentation is available at https://brianward.dev/scikit-stan/ or on ReadTheDocs (older versions and PDFs available).
Pre-compiled wheels for the package are available for MacOS, Windows, and Linux systems via pip install scikit_stan.
Source installation requires a working installation of CmdStan.
from scikit_stan import GLM
m = GLM(family='gamma') # Gamma family distribution with canonical inverse link
m.fit(X, y) # runs HMC-NUTS
predictions = m.predict(X) # generates new predictions from fitted model
score = m.score(X, y) # computes the R2 score of the fitted model on the data X and observations yThis package is licensed under the BSD 3-clause license.
It is inspired by existing packages in the Stan ecosystem like rstanarm.
This package was initially developed at the Simons Foundation by Alexey Izmailov during a summer 2022 internship under the mentorship of Brian Ward.