Causal Inference for ML
Move beyond correlation to causal inference — causal graphs, identification, and estimation. Correlation is not causation.
Level: Advanced · Category: Machine Learning · Estimated time: 7 hours
Prerequisites
- Machine Learning Basics
- Statistics for Data Science
Lessons
- Correlation vs Causation — Confounding, selection bias, and why correlation fails.
- Causal Graphs — DAGs, d-separation, and backdoor criterion.
- Do-Calculus — Interventions, do-operator, and identification.
- Propensity Score Methods — Matching, weighting, and IPW.
- Instrumental Variables — IV identification and two-stage least squares.
- Causal ML with DoWhy — Implementing causal inference in Python.
Topics covered
causal-inference, causal-graphs, do-calculus, propensity-scores