Skip to main content

Inference algorithms for models based on Luce's choice axiom.

Project description

build status code coverage documentation status

choix is a Python library that provides inference algorithms for models based on Luce’s choice axiom. These probabilistic models can be used to explain and predict outcomes of comparisons between items.

  • Pairwise comparisons: when the data consists of comparisons between two items, the model variant is usually referred to as the Bradley-Terry model. It is closely related to the Elo rating system used to rank chess players.

  • Partial rankings: when the data consists of rankings over (a subset of) the items, the model variant is usually referred to as the Plackett-Luce model.

  • Top-1 lists: another variation of the model arises when the data consists of discrete choices, i.e., we observe the selection of one item out of a subset of items.

  • Choices in a network: when the data consists of counts of the number of visits to each node in a network, the model is known as the Network Choice Model.

choix makes it easy to infer model parameters from these different types of data, using a variety of algorithms:

  • Luce Spectral Ranking

  • Minorization-Maximization

  • Rank Centrality

  • Approximate Bayesian inference with expectation propagation

Getting started

To install the latest release directly from PyPI, simply type:

pip install choix

To get started, you might want to explore one of these notebooks:

You can also find more information on the official documentation. In particular, the API reference contains a good summary of the library’s features.


Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

choix-0.3.5.tar.gz (63.1 kB view hashes)

Uploaded source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page