v0.1.1 Release

mabby is a library for simulating multi-armed bandits, with the following features:

• running simulations with different strategies on a configurable set of arms
• tracking for regret, cumulative regret, optimality, reward, and cumulative reward metrics
• visualizing the tracked metrics, allowing comparison between different strategies

mabby also currently supports and includes:

• implementations for epsilon-greedy, UCB1, and beta Thompson sampling strategies
• implementations for arms with Bernoulli and Gaussian reward distributions
• custom strategies and arms through sub-classing the abstract Strategy and Arm classes