Siraj Raval programs a virtual robot to do some house cleaning using a technique called Monte Carlo Prediction. In typical Siraj fashion he explains what it is, how it works and how to use it for reinforcement learning.