Prepared for the Very Applied Methods series at the London School of Economics and Political Science.
The material is largely based on the excellent course by Richard Morey @richarddmorey at Essex Summer School in 2011.
- Intro to the mathematical background (short, the Bayesian "mantra")
- Specifying priors
- Sampling (Metropolis-Hastings / Gibbs sampling) (also short, but it's good to have an idea what's going on under the hood).
- Convergence (extremely important but often overlooked)
Variational Inference and other approximations (used e.g. for topic models)
- Estimating "standard" models in MCMCpack
- Output checks with coda
Writing your own models in stan