In AI, sometimes, you need to plan a sequence of action that lead you to your goal. In stochastic environment, in those situation where you can’t know the outcomes of your actions, a sequence of actions is not sufficient: you need a policy.
Markov Decision Process is a mathematical framework that helps to build a policy in a stochastic environment where you know the probabilities of certain outcomes.
In this post, I give you a breif introduction of Markov Decision Process. Then, I’ll show you my implementation, in python, of the most important algorithms that can help you to find policies in stocastic enviroments. You can fine a more detailed description of the Markov Decision Process in my my slides that I’ve used for a seminar at University.