This goal of this repo is to offer some visual interpretation of bandit algorithms output.
There are currently 2 notebooks:
-
linUCB.ipynb: LinUCB algorithm (disjoint model case) from A Contextual-Bandit Approach to Personalized News Article Recommendation (Li et al 2010 )
-
bandit_model_selection.ipynb: Bandits are used for online model selection. This feature is for now experimental. It relies heavily on the Python online ML library river. This notebook might not be reproducible at the moment since this feature is not yet fully merged in the library.