In this article, I propose to implement from scratch, my own version of the Metropolis-Hastings algorithm to find parameter distributions for a dummy data example and then of a real world problem.
I figured that if I get my hands dirty, I might finally be able to understand it. I will only use numpy to implement the algorithm, and matplotlib to draw pretty things. Alternatively, scipy can be used to compute the density functions, but I will also show how to implement them using numpy.