Collapsibility of Graphical CG-Regression Models (Q4677111)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 2169671
Language Label Description Also known as
default for all languages
No label defined
    English
    Collapsibility of Graphical CG-Regression Models
    scientific article; zbMATH DE number 2169671

      Statements

      Collapsibility of Graphical CG-Regression Models (English)
      0 references
      20 May 2005
      0 references
      chain graphs
      0 references
      conditional independence graphs
      0 references
      decomposability
      0 references
      hierarchical models
      0 references
      The authors study properties of a class of models used in graphical modelling with mixed discrete and continuous data, the conditional Gaussian (CG-regression) models. These models describe the distribution of a set of response variables given a set of covariates, where both sets may comprise discrete and continuous variables. The property of collapsibility of the considered models is investigated. A model is collapsible onto a variable subset, if it can be decomposed into two simpler models, a marginal model on the subset, and a conditional model describing the dependence of the remaining variables on the subset variables. The authors obtain necessary and sufficient conditions for collapsibility for graphical CG-regression models. The implications of this result, in particular for maximum likelihood estimation, are presented and models that can be decomposed into a sequence of univariate CG-regressions are considered.
      0 references
      0 references
      David Edwards
      0 references

      Identifiers