Bayesian Statistics

Routing Summary

This folder covers comprehensive Bayesian statistics from BDA3, Statistical Rethinking, the Bayesian Workflow paper, simulation-based calibration (SBC), synthetic likelihood, and PyMC tutorials. Contains 70 notes across 8 sub-topics.

  • Need inference basics (Bayes’ theorem, conjugate priors, hierarchical)? Inference Fundamentals
  • Need model evaluation (PPC, WAIC, LOO)? Model Assessment
  • Need MCMC, HMC, or variational inference? Computation
  • Need regression, GLMs, or missing data? Regression Models
  • Need GPs, mixtures, spatial, or causal BART? Advanced Models
  • Need Bayesian causal inference (potential outcomes, BART/BCF, IV, g-computation)? Causal Inference
  • Need the iterative modeling cycle or simulation-based calibration (SBC) for validating inference algorithms? Bayesian Workflow
  • Need likelihood-free / simulation-based inference for chaotic dynamic models (synthetic likelihood)? Synthetic Likelihood

Book Overviews

Sub-topics

Sub-topicNotesDomain
Inference Fundamentals8Bayes’ theorem, conjugate models, hierarchical models (BDA3 Part I)
Model Assessment5Posterior predictive checks, model comparison, decision analysis (BDA3 Part II)
Computation5MCMC, HMC, variational inference, Stan (BDA3 Part III)
Regression Models9Bayesian regression, multilevel models, GLMs, missing data (BDA3 Part IV)
Advanced Models10GPs, mixtures, Dirichlet processes, spatial, copulas, BART, Bayesian IPW (BDA3 Part V + PyMC)
Bayesian Workflow13The iterative modeling cycle (Gelman et al. 2020) + simulation-based calibration: data-averaged posterior self-consistency, rank uniformity, the SBC algorithm, histogram diagnostics, case studies (Talts et al. 2018)
Causal Inference39Potential outcomes, BART/BCF outcome models, propensity score, IV, g-formula, metalearners, BSTS/CausalImpact, knowledge elicitation, and dynamic treatment regimes (Q-/A-learning)
Synthetic Likelihood4Likelihood-free inference for noisy chaotic dynamic models: phase-insensitive summary statistics, the MVN synthetic likelihood, MCMC exploration, Nicholson’s blowfly application (Wood 2010, Nature)

Sources

See Also