Maximum likelihood estimation in log-linear models

Abstract

We study maximum likelihood estimation in log-linear models under conditional Poisson sampling schemes. We derive necessary and sufficient conditions for existence of the maximum likelihood estimator (MLE) of the model parameters and investigate estimability of the natural and mean-value parameters under a nonexistent MLE. Our conditions focus on the role of sampling zeros in the observed table. We situate our results within the framework of extended exponential families, and we exploit the geometric properties of log-linear models. We propose algorithms for extended maximum likelihood estimation that improve and correct the existing algorithms for log-linear model analysis.

Published in at http://dx.doi.org/10.1214/12-AOS986 the Annals of Statistics (http://www.imstat.org/aos/) by the Institute of Mathematical Statistics (http://www.imstat.org)

References in corpus (7)

• On the toric algebra of graphical models
• Describing disability through individual-level mixture models for multivariate binary data
• Sequential importance sampling for multiway tables
• Closures of exponential families
• A conjugate prior for discrete hierarchical log-linear models
• Iterative proportional scaling via decomposable submodels for contingency tables
• Relations among conditional probabilities

Cited by in corpus (27)

• Nonparametric graphon estimation
• Maximum lilkelihood estimation in the $\beta$-model
• Asymptotics in directed exponential random graph models with an increasing bi-degree sequence
• Model fitting in Multiple Systems Analysis for the quantification of Modern Slavery: Classical and Bayesian approaches
• Multiple Systems Estimation for Sparse Capture Data: Inferential Challenges when there are Non-Overlapping Lists
• Degree-based network models
• Bayesian nonparametric disclosure risk estimation via mixed effects log-linear models
• Staged tree models with toric structure
• Toric invariant theory for maximum likelihood estimation in log-linear models
• Differentially Private Learning of Undirected Graphical Models using Collective Graphical Models
• Approximating faces of marginal polytopes in discrete hierarchical models
• Bootstrapping multiple systems estimates to account for model selection
• Computational information geometry: theory and practice
• $A$-Hypergeometric Distributions and Newton Polytopes
• The quantification of Simpsons paradox and other contributions to contingency table theory
• Distributed Parameter Estimation in Probabilistic Graphical Models
• Generalized Cut Polytopes for Binary Hierarchical Models
• Fitting log-linear models in sparse contingency tables using the eMLEloglin R package
• Distributed parameter estimation of discrete hierarchical models via marginal likelihoods
• On the Geometry and Extremal Properties of the Edge-Degeneracy Model
• A local approach to estimation in discrete loglinear models
• Multinomial and empirical likelihood under convex constraints: directions of recession, Fenchel duality, perturbations
• Linear and Parallel Learning of Markov Random Fields
• Longitudinal Network Models and Permutation-Uniform Markov Chains
• Iterative proportional scaling revisited: a modern optimization perspective
• The existence of maximum likelihood estimate in high-dimensional binary response generalized linear models
• On the closure of relational models