Oral exam.
This advanced course builds on the basic causal inference class by extending theory and tackling real-world data complexities with modern methods. It deepens understanding of conditional independence testing (CIT) and develops causal discovery (CD), focusing on hidden confounders, cycles, non-stationarity, multiple datasets, and high-dimensional variables. While emphasizing constraint-based CD, score-based algorithms are also covered within a broader framework. We address methodological advances, benchmarking, and inductive biases. Beyond CD as a first stage, the course studies causal effect identification, estimation under finite samples, counterfactuals, and mediation. Connections to potential outcomes, dynamic systems, and representation learning are drawn. Theory is paired with applications, proofs, and real data examples across scientific fields.
