created by Austin Poor
Using pystan
to approach a simplified version of the multi-armed bandit problem, where the goal is to use bayesian modeling to predict the probability of success for an unknown process.
The notebook bayes-bandit.ipynb has the analysis code and the file bandit.py has a 1-armed-bandit class used by the notebook.
To run the notebook using Binder, click the badge above or use this link.
There's an included docker-compose file, for running the notebook in a container, which can be run with the command:
$ docker-compose up