Introduction to Causal Graphical Models: Graphs, d-separation, do-calculus
summary
Introduces a probabilistic causal model with hidden U variables, observable Xs, deterministic structural functions and interventions modeled by replacing mechanisms (do(X=x)); uses DAGs to get the factorization P(V)=∏P(X|Parents) and conditional independence via d-separation; demonstrates cases where effect is or isn’t computable from P(V) (mediator Z can make do(X) identifiable); presents the do-calculus and aims to explain the Shpitser–Pearl ID algorithm for proving or producing formulas for identifiability (one intermediate derivation is flagged as incorrect and will be corrected).
AI-written summary · about
chapters (14)
- 0:00Introduction
- 7:15Table of Contents
- 14:38Graphs
- 17:23Causal Graphical Models
- 23:36Modified Induced Graph
- 28:57The Back Door
- 30:07Instrumental Variables
- 33:07The Big Picture
- 34:07Agenda
- 38:37Continuous Variables
- 39:07Bayesian Networks
- 41:12Graph Theory
- 45:27Bayesian Network
- 52:42Topological Ordering
related talks
01h 01mThe Mathematics of Causal Inference, with Reflections on Machine Learning and the Logic of Science
32mNew Approaches To Learning Nonparametric (Latent) Causal Graphical Models
29m🧠🧹 Causality - Mental Hygiene for Data Science
31mIntroduction to Causal Inference using pgmpy
01h 06mThe New Science of Cause and Effect
24mA Friendly Introduction To Causality
43mA Practical Guide to Causality in Python (For The Perplexed)
41mOn the Path to Causal Inference