AI 용어집
인공지능 완전 사전
Directed Acyclic Causal Diagram (DAG)
Graphical representation of causal hypotheses where nodes are variables and directed edges indicate direct influence, with no cycles possible to ensure model consistency.
Fixed Effects Model
Regression technique that controls for unobserved but time-invariant confounding variables by including indicators for each entity (individual, company, etc.).
Difference-in-Differences
Quasi-experimental strategy that compares the outcome evolution in a treatment group to that in a control group before and after intervention to estimate causal effect.
Moral Boundary
Set of variables in a DAG that, when conditioned on, open new bias paths between treatment and outcome, and should therefore be avoided in adjustment.
Treatment Heterogeneity
Variation in the causal effect of treatment between different subpopulations or individuals, often modeled to understand for whom and under what conditions an intervention is most effective.