An introduction to Markov Chain Monte Carlo using PyMC3
summary
Covers concrete MCMC practice using PyMC3 and the repo https://github.com/fonnesbeck/mcmc_pydata_london_2019 with runnable notebooks: hand-implements Metropolis, shows tuning to an acceptance ≈0.23, and then uses HMC/NUTS. Works through a radon case (log-mean ≈1.39 with P(mean>4)≈0.16 and posterior-predictive P(house>4)≈0.46), a coal-mining change-point model, and a hierarchical rat growth model, demonstrates diagnostics (R̂, ESS, autocorrelation), and the examples run and illustrate failures and fixes.
AI-written summary · about
chapters (9)
- 0:00PyMC3 tutorial setup
- 1:34Bayesian statistics benefits
- 13:52Bayesian workflow with radon
- 25:33Markov Chain Monte Carlo intro
- 41:36Metropolis-Hastings algorithm
- 51:14Building PyMC3 models
- 1:06:12Running MCMC sampling
- 1:15:49Model checking diagnostics
- 1:27:53Q&A
related talks
44mAn introduction to PyMC3
42mBayesian Non-parametric Models for Data Science using PyMC3
57mA Quickstart Guide to PyMC3
01h 26mProbabilistic Programming and Bayesian Computing with PyMC
01h 31mProbabilistic Programming and Bayesian Inference in Python
01h 27mProbabilistic Python: An Introduction to Bayesian Modeling with PyMC
34mFast, beautiful and easy Bayesian modelling: Can you have it all?
01h 27mBayesian Time Series