How can causal relationships help with prediction?
Duration
10 to 20 h
Syllabus
Part I: Bayesian networks and causal inference (“Pearl causality”)
- Introduction to conditional independence
- Introduction to graphical models
- Do calculus and deriving causal inference from Bayesian networks
- Estimation of Bayesian networks from data
- Overview of current approaches to learning causal graphs from observational data
Part II: Potential outcomes and causal inference (“Rubin causality”)
- Intro to potential outcome terminology and notation
- Causal inference with randomized experiments
- Propensity scores and matching
Prerequisites
Solid background in Statistics/Econometrics
Credential
Forms part of Summer School Week II, part of Statistical Modelling and Inference MSc course, also taught in some Data Science Center workshops