Markovian acyclic directed mixed graphs for discrete data

DOI10.1214/14-AOS1206zbMATH Open1302.62148arXiv1301.6624OpenAlexW2114089531MaRDI QIDQ464191FDOQ464191

Thomas S. Richardson, Robin J. Evans

Publication date: 17 October 2014

Published in: The Annals of Statistics (Search for Journal in Brave)

Abstract: Acyclic directed mixed graphs (ADMGs) are graphs that contain directed ($ightarrow$) and bidirected ($leftrightarrow$) edges, subject to the constraint that there are no cycles of directed edges. Such graphs may be used to represent the conditional independence structure induced by a DAG model containing hidden variables on its observed margin. The Markovian model associated with an ADMG is simply the set of distributions obeying the global Markov property, given via a simple path criterion (m-separation). We first present a factorization criterion characterizing the Markovian model that generalizes the well-known recursive factorization for DAGs. For the case of finite discrete random variables, we also provide a parameterization of the model in terms of simple conditional probabilities, and characterize its variation dependence. We show that the induced models are smooth. Consequently, Markovian ADMG models for discrete variables are curved exponential families of distributions.

Full work available at URL: https://arxiv.org/abs/1301.6624

Recommendations

• Markov Properties for Acyclic Directed Mixed Graphs
• Markov properties for mixed graphs
• The hidden life of latent variables: Bayesian learning with mixed graph models
• On the Markov Equivalence of Chain Graphs, Undirected Graphs, and Acyclic Digraphs
• A characterization of Markov equivalence classes for acyclic digraphs

zbMATH Keywords

conditional independencegraphical modelparameterizationcurved exponential familyacyclic directed mixed graphm-separation

Mathematics Subject Classification ID

Multivariate analysis (62H99) Directed graphs (digraphs), tournaments (05C20) Edge subsets with special properties (factorization, matching, partitioning, covering and packing, etc.) (05C70)

Cites Work

• Causation, prediction, and search
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Causality. Models, reasoning, and inference
• Ancestral graph Markov models.
• Title not available (Why is that?)
• Probability distributions with summary graph structure
• Causal diagrams for empirical research
• Markov Properties for Acyclic Directed Mixed Graphs
• Marginal models for categorical data
• The hidden life of latent variables: Bayesian learning with mixed graph models
• Title not available (Why is that?)
• Identifying the consequences of dynamic treatment strategies: a decision-theoretic overview
• Stable mixed graphs
• Alternative Markov properties for chain graphs
• Discrete chain graph models
• Markov properties for mixed graphs
• Smoothness of Gaussian conditional independence models

Cited In (13)

• Title not available (Why is that?)
• Towards standard imsets for maximal ancestral graphs
• Conditional independence collapsibility for acyclic directed mixed graph models
• Smooth, identifiable supermodels of discrete DAG models with latent variables
• Compartmental model diagrams as causal representations in relation to DAGs
• Margins of discrete Bayesian networks
• Graphs for margins of Bayesian networks
• Marginalization and conditioning for LWF chain graphs
• Marginal Markov subgraph of Bayesian network and its applications
• Markov equivalence of marginalized local independence graphs
• Markov Properties for Acyclic Directed Mixed Graphs
• Nested covariance determinants and restricted trek separation in Gaussian graphical models
• A decomposition-based algorithm for learning the structure of multivariate regression chain graphs

Uses Software

• TETRAD

This page was built for publication: Markovian acyclic directed mixed graphs for discrete data

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q464191)