Description: (Same as Computer Science M262C.) Lecture, four hours. Requisite: one graduate probability or statistics course such as course 200B, 202B, or Computer Science 262A. Review of Bayesian networks, causal Bayesian networks, and structural equations. Learning causal structures from data. Identifying causal effects. Covariate selection and instrumental variables in linear and nonparametric models. Simpson paradox and confounding control. Logic and algorithmization of counterfactuals. Probabilities of counterfactuals. Direct and indirect effects. Probabilities of causation. Identifying causes of events. Letter grading.

Units: 4.0