Explore the Hamiltonian Monte Carlo revolution and its impact on Bayesian statistical computation in this 43-minute conference talk. Dive into the world of probabilistic programming with PyMC3, an open-source Python package supported by NumFOCUS. Learn how recent advancements have enabled effective computation for complex models, making them accessible to programmers and statisticians. Discover practical applications through examples in invasion statistics and basketball analytics. Gain insights into the Monte Carlo problem, Bayesian inference, and the implementation of Hamiltonian Monte Carlo in PyMC3. No prior knowledge of Bayesian statistics is required, though basic Python skills will be beneficial.
The Hamiltonian Monte Carlo Revolution Is Open Source - Probabilistic Programming with PyMC3
Overview
Syllabus
Introduction
Welcome
Who am I
Download Jupyter Notebook
Download Jupyter Container
Kibo
Kibo Careers
Motivation Examples
Invasion Statistics
Basketball Analytics
Drawing Fouls
Basketball
What is probabilistic programming
What is interesting about probabilistic programming
The Monte Carlo Problem
Using PyMC3
The Game Show
Adding Data
Question
New Bugs
Documentation
Case Study
Last 2 Minute Report
The Model
The Season Factor
Metropolis Hastings
Bayesian Inference
The Curse of Dimensionality
Hamiltonian Monte Carlo
How PyMC3 implements Hamiltonian Monte Carlo
No Uturn Sampler
Good Mixing
Foul Call Rate
Books
Stan
Thank you
Resources
Taught by
Open Data Science
