In this talk I will show you how to do Bayesian Inference via fully-worked examples using game data from baseball and soccer. By comparing predictions made by plugging in Bayesian estimates to those made by plugging in maximum likelihood estimates, we see how critical it is that your predictions reflect not just your estimate of the most likely outcome, but also your degree of uncertainty.
Takeaways from this talk:
posterior predictive checks - do the model predictions make sense?
Introduction to estimation by Markov-Chain Monte Carlo (MCMC) simulation, using Stan (mc-stan.org), a probabilistic programming platform which does full Bayesian inference using Hamiltonian Monte Carlo (HMC).
Ability to discriminate between Bayesian and point-wise estimation methods, regardless of what they are called.