bayescart documentation

bayescart is a Python package for Bayesian Classification and Regression Trees (CART) posterior sampling using custom, advanced tempering methods.

This package provides classes and functions to build, sample, and evaluate Bayesian classification and regression trees using Markov chain Monte Carlo (MCMC) methods. It supports various tempering strategies (geometric, likelihood-based, and pseudo-prior) to improve mixing in multi-modal posterior distributions.

For more information on the creation of the package, see this dedicated page.

For theoretical background on Bayesian CART, and the specific tempering strategies implemented in this package, check this detailed blog series.

For an example on how to use this package, see this tutorial notebook.

Contents:

• bayescart
  • bayescart package
    • Submodules
    • bayescart.bcart module
    • bayescart.eval module
    • bayescart.exceptions module
    • bayescart.mytyping module
    • bayescart.node module
    • bayescart.node_data module
    • bayescart.parallel_tempering_base module
    • bayescart.tempering_geom_lik module
    • bayescart.tempering_geometric module
    • bayescart.tempering_pseudo_prior module
    • bayescart.tree module
    • bayescart.utils module
    • Module contents