DataTerminology.comNEW
Also known as: DAG
A directed graph with no directed cycles, often used to represent causal structure or dependencies.
DESCRIPTION
A Directed Acyclic Graph (DAG) is a finite directed graph with no cycles, enabling topological ordering of nodes. In causal inference, DAGs encode assumptions about cause–effect relations and guide valid adjustment sets. In computing, DAGs model dependencies in workflows, compilers, and version control. Their acyclic property simplifies reasoning over orderings and reachability.
ALTERNATIVE DEFINITIONS
Wikipedia
In mathematics and computer science, a directed acyclic graph (DAG) is a directed graph with no directed cycles.
Concise formal definition focusing on the absence of cycles.
Wolfram MathWorld
A directed acyclic graph is a directed graph containing no directed cycles.
Matches the canonical mathematical definition from a reference site.
RELATED TERMS
Causal Graph Bayesian Network Markov Random Field Topological Ordering d-separation Do-Calculus Structural Causal Model Graph Theory Graph Neural Network Workflow DAG
COMPETING TERMS
Undirected Graph
RELATIONS
- Broader: Graph
- Narrower: Bayesian Network
KNOWLEDGE AREAS
Mathematics Data Science AI Data Engineering
TARGET AUDIENCE
Data Scientist Data Engineer AI Engineer Data Architect Information Analyst
EXTERNAL SOURCES
- Wikipedia - Directed acyclic graph
- Wolfram MathWorld - Reference entry with properties
- Stanford CS228 - Bayesian networks on DAGs
- Apache Airflow - DAG concept in workflow orchestration
TOP QUESTIONS
What is a Directed Acyclic Graph (DAG)?
A DAG is a directed graph without directed cycles, enabling a topological ordering of nodes.
Category: Definition
Why are DAGs used in causal inference?
They encode assumptions about causal relationships and support valid adjustment via d-separation and related criteria.
Category: Example
How are DAGs applied in data engineering?
Workflow tools like Airflow use DAGs to schedule tasks respecting dependencies and avoid cycles.
Category: Application
What is a topological order in a DAG?
An ordering of nodes such that every directed edge goes from earlier to later in the order.
Category: Definition
What is a common mistake when using DAGs for causality?
Including cycles or adjusting for colliders, which invalidates causal interpretations or introduces bias.
Category: Mistakes
Can DAGs represent feedback loops?
No. Feedback implies cycles; DAGs exclude cycles by definition. Alternate models are needed for feedback.
Category: Misunderstanding
What is d-separation in a DAG?
A graphical criterion to determine conditional independence between sets of nodes.
Category: Definition
Are DAGs unique for a given problem?
No. Multiple DAGs may encode the same conditional independencies; domain knowledge guides selection.
Category: Opinion
Where to learn more about DAGs?
Bayesian network resources (e.g., Stanford CS228) and causal inference texts provide formal introductions and examples.
Category: Learning