This is an OCaml library that provides an implementation of the epsilon-greedy solution to the multiarmed-bandit problem, using exponential moving averages for a faster adaptation to new trends.
Practical use: finding out which variant of a user interface is most often successful while maximizing the overall success rate as the system is learning.
- 20 lines of code that will beat A/B testing every time
- Multi-armed bandit problem on Wikipedia
- Exponential smoothing on Wikipedia
$ make $ make install
$ make uninstall