Estimating Probabilities with Bayesian Modeling in Python
Will Koehrsen

Great article — and very nicely demonstrates how awesome Bayesian methods are on very small samples. Moreover, when I look around, most of my datasets follow an exponential distribution — the massive majority are tiny, whereas only a few are large enough to get good results out of a very deep neural network. I can’t recommend highly enough to other data science practitioners, set the neural networks aside for a couple of evenings — and have a play with statistical sampling: I guarantee you’ll have a massive ‘oh my God — this is incredible!’ moment — and get some amazing insights into the small datasets that you didn’t dare touch because normal ML methods would overfit.