bayesianbandits

Bayesian multi-armed bandits with conjugate online learning. Sometimes, you need to know your decisions are close to optimal at all times.

Getting started

New to bandits? The introduction explains when you’d use one and how to choose your setup. The Quick Start walks through a complete example in 5 minutes.

How-to guides

Task-oriented recipes: pipelines, decay, reward functions, delayed rewards, production deployment, sparse features.

Explanation

Why this library makes the choices it does: Explanation.

Mathematical reference

Update equations, hyperparameter semantics, and departures from textbook for each estimator family: Mathematical Reference.

API reference

Full details on every class and method: API Reference.