Maximum likelihood estimation in log-linear models

DOI10.1214/12-AOS986zbMATH Open1274.62389arXiv1104.3618OpenAlexW2095183921MaRDI QIDQ447845FDOQ447845

Authors: Stephen E. Fienberg, Alessandro Rinaldo 

Publication date: 29 August 2012

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

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.

Full work available at URL: https://arxiv.org/abs/1104.3618

Recommendations

• scientific article; zbMATH DE number 4094597
• Maximum Likelihood Methods for Log-Linear Models When Expected Frequencies are Subjected to Linear Constraints
• scientific article; zbMATH DE number 3984350
• On maximum likelihood estimation in sparse contingency tables
• Maximum likelihood estimation for the negative multinomial log-linear model

zbMATH Keywords

log-linear modelsNewton-Raphson algorithmextended exponential familiesextended maximum likelihood estimatorssampling zeros

Mathematics Subject Classification ID

Parametric inference (62F99) Contingency tables (62H17)

Cites Work

• polymake: a framework for analyzing convex polytopes
• Title not available (Why is that?)
• Title not available (Why is that?)
• A conjugate prior for discrete hierarchical log-linear models
• Three centuries of categorical data analysis: Log-linear models and maximum likelihood estima\-tion
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• The log-linear group-lasso estimator and its asymptotic properties
• Maximum likelihood estimation in log-linear models
• The mode oriented stochastic search (MOSS) algorithm for log-linear models with conjugate priors
• On the geometry of discrete exponential families with application to exponential random graph models
• Title not available (Why is that?)
• On the toric algebra of graphical models
• Title not available (Why is that?)
• Convex Analysis
• Algebraic Statistics for Computational Biology
• Title not available (Why is that?)
• Lectures on algebraic statistics
• Introduction to Toric Varieties. (AM-131)
• Lectures on Polytopes
• Title not available (Why is that?)
• Maximum likelihood estimation in the \(\beta\)-model
• Generalized Iterative Scaling for Log-Linear Models
• Goodness-of-fit statistics for discrete multivariate data
• I-divergence geometry of probability distributions and minimization problems
• Information projections revisited
• Lectures on natural exponential families and their variance functions
• Title not available (Why is that?)
• Describing disability through individual-level mixture models for multivariate binary data
• Central limit theorems for multinomial sums
• Prior induction in log-linear models for general contingency table analysis.
• Multinomial-Poisson homogeneous models for contingency tables.
• Goodness-of-Fit Tests for Log-Linear Models in Sparse Contingency Tables
• Relations among conditional probabilities
• Likelihood inference in exponential families and directions of recession
• A geometric interpretation of Darroch and Ratcliff's generalized iterative scaling
• Homogeneous Linear Predictor Models for Contingency Tables
• Existence of MLEs for discrete linear exponential models
• Log-linear models and frequency tables with small expected cell counts
• Closures of exponential families
• Polyhedral conditions for the nonexistence of the MLE for hierarchical log-linear models
• Generalized maximum likelihood estimates for exponential families
• Algebraic statistics and contingency table problems: log-linear models, likelihood estimatio, and disclosure limitation
• The compactification of generalized linear models
• Convex cores of measures on R d
• Bayesian inference for Poisson and multinomial log-linear models

Cited In (53)

• Longitudinal network models and permutation‐uniform Markov chains
• Maximum likelihood estimation in the \(\beta\)-model
• Computationally efficient likelihood inference in exponential families when the maximum likelihood estimator does not exist
• Conditional inference of Poisson models and information geometry: an ancillary review
• Title not available (Why is that?)
• Title not available (Why is that?)
• On maximum likelihood estimation for log-linear models with non-ignorable non-response
• MAXIMUM LIKELIHOOD ESTIMATION OF GROSS FLOWS
• A note on A. Albert and J. A. Anderson's conditions for the existence of maximum likelihood estimates in logistic regression models
• Statistical inference in a directed network model with covariates
• A scheme for molecular computation of maximum likelihood estimators for log-linear models
• Improving the power of hypothesis tests in sparse contingency tables
• Approximate maximum likelihood estimation of the autologistic model
• On the principle of monotone likelihood and log-linear models
• The maximum likelihood estimator for joint mean and dispersion in generalized linear models of the lognormal distribution
• Accurate directional inference for vector parameters in linear exponential families
• Computing with Fisher geodesics and extended exponential families
• On maximum likelihood estimation in sparse contingency tables
• Bootstrapping the maximum likelihood estimator in high-dimensional log- linear models
• A likelihood ratio test of quasi-independence for sparse two-way contingency tables
• Maximum likelihood estimators of the parameters of the log-logistic distribution
• Staged tree models with toric structure
• Assessing Bayesian Semi‐Parametric Log‐Linear Models: An Application to Disclosure Risk Estimation
• Title not available (Why is that?)
• Maximum Likelihood Methods for Log-Linear Models When Expected Frequencies are Subjected to Linear Constraints
• Bootstrapping multiple systems estimates to account for model selection
• Generalized cut polytopes for binary hierarchical models
• The existence of maximum likelihood estimate in high-dimensional binary response generalized linear models
• On the role of the overall effect in exponential families
• \(A\)-hypergeometric distributions and Newton polytopes
• Log-linear model estimation for stratified educational data
• Loglinear model selection and human mobility
• Directed weighted random graphs with an increasing bi-degree sequence
• Bayesian nonparametric disclosure risk estimation via mixed effects log-linear models
• Confidence intervals for discrete log-linear models when MLE does not exist
• On the closure of relational models
• Multiple Systems Estimation for Sparse Capture Data: Inferential Challenges When There Are Nonoverlapping Lists
• Asymptotics in directed exponential random graph models with an increasing bi-degree sequence
• An approximation to maximum likelihood estimates in reduced models
• Local conditional and marginal approach to parameter estimation in discrete graphical models
• Maximum likelihood estimation for multinomial-Poisson models: a generalization of Birch's numerical invariance results
• Maximum likelihood estimation in log-linear models
• Toric invariant theory for maximum likelihood estimation in log-linear models
• Iterative scaling in curved exponential families
• Polyhedral conditions for the nonexistence of the MLE for hierarchical log-linear models
• The Asymptotic Covariance Matrix of the Maximum Likelihood Parameter Estimator in Conditional Poisson Log-linear Models
• When are first-order asymptotics adequate? A diagnostic
• New improved estimators for overdispersion in models with clustered multinomial data and unequal cluster sizes
• Maximum likelihood estimation for discrete exponential families and random graphs
• Title not available (Why is that?)
• Title not available (Why is that?)
• Iterative Proportional Scaling Revisited: A Modern Optimization Perspective
• Title not available (Why is that?)

Uses Software

• R
• polymake

This page was built for publication: Maximum likelihood estimation in log-linear models

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q447845)