Introduction to Causal Inference
summary
Covers practical steps: uses Jupyter notebooks to run g-computation on the LaLonde job‑training data and estimates a treatment effect near $17,000; uses the causal-curve python package to fit continuous-treatment curves for blood‑lead vs. test scores and finds roughly 20% mediation by income. Explains how confounders, colliders and mediators change estimates, lists assumptions (temporality, positivity, ignorability), and shows the models and code actually run so claimed effects are reproducible rather than just conceptual.
AI-written summary · about
chapters (33)
- 0:00Causal inference intro
- 0:23Why causal inference matters
- 1:26Vitamin D to COVID-19
- 3:11Experiments not always possible
- 4:36Causal inference middle ground
- 5:39Causal vs ML questions
- 6:42Causal graphs primer
- 8:04Three critical causal relationships
- 9:06Confounders distort relationships
- 10:52Ice cream and crime case study
- 12:17Controlling confounders in practice
- 13:20Colliders to avoid
- 14:44Mediators to ignore
- 16:29Reality needs domain knowledge
- 17:54Structure learning controversial
- 20:39Live demo in notebooks
- 26:15Control confounders not mediators
- 26:35Variable importance fails
- 30:09Four key assumptions
- 32:33Testing assumptions with examples
- 35:38G computation technique
- 40:11G computation step-by-step
- 44:24Handling colliders in G
- 46:52G vs other methods
- 47:16Causal curves for treatments
- 50:07Lead exposure and test scores
- 53:13Mediation effects
- 54:37First steps flowchart
- 55:43Domain knowledge in industry
- 58:09Avoid p-value fishing
- 1:01:42Randomization gold standard
- 1:05:11Bayesian causal framework
- 1:06:26Q&A
related talks
24mAn introduction to Causal Inference with Python – making accurate estimates of cause and effect from
31mIntroduction to Causal Inference using pgmpy
06mCausal Machine Learning – What’s in it for Data Scientists?
05mBackstage interview
26m"Causal Discovery in Python"
37mDid Your Rollout Actually Work?
35mMental Models to Use and Avoid as a Data Scientist
01h 24mIntroductory tutorial on data exploration and statistical models