Causal Effects via the Do-operator | Overview & Example

🤝 Want your team maximizing Claude? I run 1:1 and team AI workshops for companies doing $1M+ per year: https://aibuilder.academy/yt/dejZzJIZdow This is the 3rd video in a series on causal effects. Here I discuss a new way to formulate the average treatment effect (ATE) using the do-operator. This alternative formulation unlocks new paths toward estimating causal effects from observational data. Series Playlist: https://www.youtube.com/playlist?list=PLz-ep5RbHosVVTz9HEzpI4d6xpWsc8rOa 📰 Read more: https://medium.com/towards-data-science/causal-effects-via-the-do-operator-5415aefc834a?sk=67a99b46a5ff8821cfb1443aa351be0e Resources: - An Introduction to Causal Inference by Judea Pearl: https://www.degruyter.com/document/doi/10.2202/1557-4679.1203/html - An Introduction to Propensity Score Methods for Reducing the Effects of Confounding in Observational Studies by Peter Austin: https://www.tandfonline.com/doi/full/10.1080/00273171.2011.568786 Introduction - 0:00 Observational vs Interventional Data - 0:35 2 Formulations of ATE - 2:23 do-operator - 5:26 Identifiability - 7:05 Truncated Factorization Formula - 10:34 Coping with Unmeasured Confounders - 10:52 Interventional Distribution via Parents - 12:34 Key Points - 13:08