Atlas · GenAI 2026

Causal Inference

Causal inference (DoWhy, EconML)

conceptPeak: 2021Causal InferenceAI consensus: 2/3

Prerequisites

  • Causal inference methods build directly on statistical estimation, hypothesis testing, and regression to move from correlation to causation

  • Treatment effect estimation, instrumental variables, and propensity scores are extensions of regression frameworks

  • Causal inference often addresses cases where controlled experiments are impossible — understanding what A/B tests do helps understand what causal inference replaces

Recommended reference

Cunningham, S. (2021) Causal Inference: The Mixtape — free online; best accessible intro; Hernan & Robins 'What If' for deeper treatment

Notes from AI deep research

Anthropic Opus

DoWhy i EconML: z 'co koreluje' na 'co powoduje'. W 2026 kluczowe dla A/B testow LLM systemow i policy evaluation

Google Deep Think

Przechodzenie od korelacji do przyczynowości [G#4]

Related skills