For Bayesian Machine Learning with Generative AI Applications, MS in Applied Data Science, The University of Chicago. Each demo runs in your browser, with nothing to install.
Monte Carlo · the posterior predictive Back to Our Evil World Draw θ from the posterior Beta(3, 9), then the next 10 days from Bin(10, θ); again, and again, and watch the histogram settle on the exact answer. Markov chain Monte Carlo · the Metropolis algorithm Metropolis on Two Humps Propose, then accept or reject, one step at a time or five thousand at speed, with the proposal's width as the knob: too narrow and the walk crawls, too wide and it rarely moves.