Bayesian target encoding

bayte offers a lightweight, scikit-learn-compliant[1] implementation of Bayesian Target Encoding. The algorithm was introduced in 2019 by Slakey et al.[2], with ensemble modeling methodology from Larionov[3]. Our explanation of the algorithm is available here.

Installation

To install bayte from PyPI, run

$ python -m pip install bayte

This is the preferred method to install bayte.

Contents

Quickstart

• When should you use this package?
  • Encoder comparison
  • Ensemble methodology
  • Comparative encoding methodology
  • Modeling algorithms
  • Datasets
  • Performance evaluation

How-to guides

• Encode your categorical variables
  • Basic usage
  • Changing hyperparameter initialization
• Build a model through repeated encoding

Explanation

• Encoding algorithm
  • Refresher on bayesian statistics
  • Procedure

Reference

• GitHub repository

[1]

Lars Buitinck, Gilles Louppe, Mathieu Blondel, Fabian Pedregosa, Andreas Mueller, Olivier Grisel, Vlad Niculae, Peter Prettenhofer, Alexandre Gramfort, Jaques Grobler, Robert Layton, Jake VanderPlas, Arnaud Joly, Brian Holt, and Gaël Varoquaux. API design for machine learning software: experiences from the scikit-learn project. In ECML PKDD Workshop: Languages for Data Mining and Machine Learning, 108–122. 2013.

[2]

Austin Slakey, Daniel Salas, and Yoni Schamroth. Encoding categorical variables with conjugate bayesian models for wework lead scoring engine. CoRR, 2019. URL: http://arxiv.org/abs/1904.13001, arXiv:1904.13001.

[3]

Michael Larionov. Sampling techniques in bayesian target encoding. CoRR, 2020. URL: https://arxiv.org/abs/2006.01317, arXiv:2006.01317.