AI-ordlista
Den kompletta ordlistan över AI
Causal Structural Model (CSM)
Mathematical representation formalizing causal relationships between variables through a system of structural equations and a causal graph. This framework allows distinguishing correlation from causation and calculating the effects of interventions.
Directed Acyclic Graph (DAG)
Graphical structure where nodes represent variables and directed edges represent direct causal relationships, without cycles. DAGs are fundamental for representing causal assumptions and applying identification criteria.
Simultaneous Structural Equations (SSE)
System of equations where each endogenous variable is expressed as a function of its direct causes and an error term. SSEs quantify causal mechanisms and allow simulating the effects of interventions.
do() interventions
Mathematical operator representing an external intervention on a variable, modifying its value independently of its normal causes. do() interventions allow distinguishing conditional probabilities from causal probabilities.
Causal Probability
Probability distribution resulting from an external intervention rather than passive observation. It quantifies the consequences of actions and fundamentally differs from standard conditional probabilities.
Confounding Variables
Variables that simultaneously influence both cause and effect, creating spurious associations in observational data. Their identification and adjustment are crucial for valid causal inference.
Counterfactuals
Questions about what would have happened if a different decision had been made, given the current state of the world. Counterfactuals are essential for causal reasoning and explaining phenomena.
Causal Mediation
Analysis decomposing the total effect of a treatment into direct and indirect effects passing through mediators. Causal mediation allows understanding the underlying mechanisms of causal relationships.
Total Causal Effect
Global impact of an intervention on a target variable, including all direct and indirect causal paths. The total effect captures the complete influence of one variable on another in the system.
Direct Causal Effect
Influence of one variable on another that is not mediated by intermediate variables in the model. The direct effect represents the immediate causal mechanism between two variables.
Indirect Causal Effect
Influence of one variable on another transmitted through one or more mediators. Indirect effects reveal complex causal chains and underlying mechanisms.
Causal Diagram
Visual representation of causal hypotheses using nodes and directed edges to encode knowledge about causal mechanisms. Causal diagrams facilitate reasoning and bias identification.