Glossary · Marketing Engineering

Causal Discovery

also: PC algorithm · constraint-based causal discovery

Definition

Causal discovery is the family of algorithms that infer directed causal structure from observational data alone, using conditional-independence tests (PC, FCI) or score-based search (GES). Applied to business data it produces testable DAGs that replace ad-hoc causal intuitions with falsifiable hypotheses.

Unlike econometric methods that estimate effects given an assumed causal graph, causal discovery learns the graph itself. The PC algorithm (Spirtes, Glymour, Scheines) iteratively tests conditional independence between variable pairs; edges are removed when a separating set is found, orientation follows from collider patterns. FCI relaxes the causal-sufficiency assumption, handling latent confounders. Output is a DAG (or PAG) that constrains which interventions are identifiable and which require experiments.

Essays on this concept