Skip to main content

Python Multi-Armed Bandit Library

Project description

PyBandits

GitHub Actions Workflow Status PyPI - Version PyPI - Python Version alt text

PyBandits is a Python library for Multi-Armed Bandit. It provides an implementation of stochastic Multi-Armed Bandit (sMAB) and contextual Multi-Armed Bandit (cMAB) based on Thompson Sampling.

For the sMAB, we implemented a Bernoulli multi-armed bandit based on Thompson Sampling algorithm Agrawal and Goyal, 2012. If context information is available we provide a generalisation of Thompson Sampling for cMAB Agrawal and Goyal, 2014 implemented with PyMC3, an open source probabilistic programming framework for automatic Bayesian inference on user-defined probabilistic models.

Installation

This library is distributed on PyPI and can be installed with pip. The latest release is version 0.0.2. pybandits requires a Python version >= 3.8.

pip install pybandits

The command above will automatically install all the dependencies listed in requirements.txt. Please visit the installation page for more details.

Getting started

A short example, illustrating it use. Use the sMAB model to predict actions and update the model based on rewards from the environment.

import numpy as np
import random
from pybandits.core.smab import Smab

# init stochastic Multi-Armed Bandit model
smab = Smab(action_ids=['Action A', 'Action B', 'Action C'])

# predict actions
pred_actions, _ = smab.predict(n_samples=100)

n_successes, n_failures = {}, {}
for a in set(pred_actions):

    # simulate rewards from environment
    n_successes[a] = random.randint(0, pred_actions.count(a))
    n_failures[a] = pred_actions.count(a) - n_successes[a]

    # update model
    smab.update(action_id=a, n_successes=n_successes[a], n_failures=n_failures[a])

Documentation

For more information please read the full documentation and tutorials.

Info for developers

PyBandits is supported by the AI for gaming and entertainment apps community.

The source code of the project is available on GitHub.

git clone https://github.com/playtikaoss/pybandits.git

You can install the library and the dependencies with one of the following commands:

pip install .                        # install library + dependencies
pip install .[develop]               # install library + dependencies + developer-dependencies
pip install -r requirements.txt      # install dependencies
pip install -r requirements-dev.txt  # install developer-dependencies

As suggested by the authors of pymc3 and pandoc, we highly recommend to install these dependencies with conda:

conda install -c conda-forge pandoc
conda install -c conda-forge pymc3

To create the file pybandits.whl for the installation with pip run the following command:

python setup.py sdist bdist_wheel

To create the HTML documentation run the following commands:

cd docs
make html

Run tests

Tests can be executed with pytest running the following commands. Make sure to have the library installed before to run any tests.

cd tests
pytest -vv                                      # run all tests
pytest -vv test_testmodule.py                   # run all tests within a module
pytest -vv test_testmodule.py -k test_testname  # run only 1 test
pytest -vv -k 'not time'                        # run all tests but not exec time

License

MIT License

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

pybandits-1.1.0.tar.gz (36.1 kB view details)

Uploaded Source

Built Distribution

pybandits-1.1.0-py3-none-any.whl (48.9 kB view details)

Uploaded Python 3

File details

Details for the file pybandits-1.1.0.tar.gz.

File metadata

  • Download URL: pybandits-1.1.0.tar.gz
  • Upload date:
  • Size: 36.1 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/1.8.4 CPython/3.8.18 Linux/6.5.0-1025-azure

File hashes

Hashes for pybandits-1.1.0.tar.gz
Algorithm Hash digest
SHA256 16bdb5df6f6ebb98a44dabf0d1a219825c697d64f3deacece525e64949b7731c
MD5 2fbae4cbb1213bd17e5959650cbed8c3
BLAKE2b-256 e921b88b09f2091dc664ced459ce7eafe58197a0263614142d614a5b3eec4fe5

See more details on using hashes here.

File details

Details for the file pybandits-1.1.0-py3-none-any.whl.

File metadata

  • Download URL: pybandits-1.1.0-py3-none-any.whl
  • Upload date:
  • Size: 48.9 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/1.8.4 CPython/3.8.18 Linux/6.5.0-1025-azure

File hashes

Hashes for pybandits-1.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 79625cd50137ef44708ba1a1a52edba4a28c1155c496c09980f35089917d8b7a
MD5 1814ff1a5cba00e12093205f8996aa87
BLAKE2b-256 38d57369b0c2daab5c2d92eeac1dd2f4f1e1ff5580ae4725397e51a9478d77ee

See more details on using hashes here.

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