Probabilistic Programming and Bayesian Computing with PyMC
summary
Uses PyMC v5 (with an optional JAX NUTS backend) to fit a marketing regression, shows MCMC diagnostics (4 chains, thousands of draws, warnings from divergences) and improves a useless linear fit by adding 14 Fourier basis with Laplace priors, a Student‑t likelihood, adstock + logistic saturation on spend, and finally a Gaussian random walk time‑varying coefficient. The final time‑varying model fits posterior predictive checks substantially better, and pyMarketing wraps this whole pipeline into a single fit call.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:06Bayesian statistics fundamentals
- 12:17PyMC regression model setup
- 25:23MCMC sampling in PyMC
- 32:48Validating model results
- 46:52Adding domain knowledge
- 50:47Seasonality and trend
- 1:01:46Advanced posterior sampling
- 1:17:22Time-varying marketing parameters
- 1:24:44Model comparison
related talks
42mSolving Real-World Business Problems with Bayesian Modeling
41mPyMC-Marketing: Customer and Marketing Analytics the Easy Way
01h 27mProbabilistic Python: An Introduction to Bayesian Modeling with PyMC
42mBayesian Non-parametric Models for Data Science using PyMC3
43mProbabilistic Programming in the Real World
39mProbablistic Programming Data Science with PyMC3
01h 27mBayesian Time Series
01h 31mProbabilistic Programming and Bayesian Inference in Python