tsampling

Thompson Sampling Multi-Armed Bandit for Python

A package that implements the Thompson Sampling approach for a Multi-Armed Bandit. The purpose of this project is to help people simply create and maintain Thompson Sampling experiments that have Bernoulli and Poisson distributions.

Usage

Setting up the experiment:

The following method will instantiate the experiment with the default priors.

from tsampling.bernoulli import BernoulliExperiment

experiment = BernoulliExperiment(arms=2)

You can also set your custom priors by using the Priors module:

from tsampling.bernoulli import BernoulliExperiment
from tsampling.priors import BetaPrior

pr = BetaPrior()
pr.add_one(mean=0.5, variance=0.2, effective_size=10, label="option1")
pr.add_one(mean=0.6, variance=0.3, effective_size=30, label="option2")
experiment = BernoulliExperiment(priors=pr)

Performing an action:

You can randomly choos which arm to "pull" in the multi-armed bandit:

experiment.choose_arm()

Updating reward:

You can update the different arms information by adding reward information:

rewards = [{"label":"option1", "reward":1}, {"label":"option2", "reward":0}]
experiment.add_rewards(rewards)

Plotting Posterior Distribution:

You can plot the posterior distribution

experiment.plot_posterior()

Installation

Pip

pip install tsampling

License

Free software: MIT license

Credits

This package was created with Cookiecutter_ and the audreyr/cookiecutter-pypackage_ project template.

_Cookiecutter: https://github.com/audreyr/cookiecutter audreyr/cookiecutter-pypackage: https://github.com/audreyr/cookiecutter-pypackage