Monte Carlo simulation of A/B tests with blended traffic mix
This is what I call "academic code". I hacked it together to help me understand a problem I was working on. It has very limited error-handling, isn't robust, isn't very parameterised etc.
Sorry about that. If you'd like to help clean it up, I welcome pull requests.