Iterative scaling in curved exponential families

DOI10.1111/SJOS.12139MaRDI QIDQ2949880zbMATH OpenOpenAlexFDO

Authors Anna Klimova, Tamás Rudas

Publication date 5 October 2015

Published in Scandinavian Journal of Statistics (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1307.3282

zbMATH Keywords

contingency tablesBregman divergenceiterative proportional fittingcurved exponential familyrelational modelmaximum likelihood estimateoverall effectgeneralized odds ratio

Mathematics Subject Classification ID

Asymptotic properties of parametric estimators (62F12) Estimation in multivariate analysis (62H12) Contingency tables (62H17)

Abstract: The paper describes a generalized iterative proportional fitting procedure which can be used for maximum likelihood estimation in a special class of the general log-linear model. The models in this class, called relational, apply to multivariate discrete sample spaces which do not necessarily have a Cartesian product structure and may not contain an overall effect. When applied to the cell probabilities, the models without the overall effect are curved exponential families and the values of the sufficient statistics are reproduced by the MLE only up to a constant of proportionality. The paper shows that Iterative Proportional Fitting, Generalized Iterative Scaling and Improved Iterative Scaling, fail to work for such models. The algorithm proposed here is based on iterated Bregman projections. As a by-product, estimates of the multiplicative parameters are also obtained.

Recommendations

• A modified iterative proportional scaling algorithm for estimation in regular exponential families
• Unified generalized iterative scaling and its applications
• Iterative proportional fitting for nonhierarchical log-linear models
• Iterative proportional scaling via decomposable submodels for contingency tables
• A fast algorithm for iterative proportional fitting in log-linear models

Cites work

• scientific article; zbMATH DE number 1817585 (Why is no real title available?)
• scientific article; zbMATH DE number 3157499 (Why is no real title available?)
• scientific article; zbMATH DE number 2107836 (Why is no real title available?)
• scientific article; zbMATH DE number 3296905 (Why is no real title available?)
• Coordinate-free analysis of trends in British social mobility
• Generalized Iterative Scaling for Log-Linear Models
• I-divergence geometry of probability distributions and minimization problems
• Iterative scaling and coordinate descent methods for maximum entropy models
• Maximum likelihood estimation in log-linear models
• On a Least Squares Adjustment of a Sampled Frequency Table When the Expected Marginal Totals are Known
• Relational models for contingency tables
• Two algorithms for fitting constrained marginal models

Cited in

(13)

• Rényi 100, quantitative and qualitative (in)dependence
• Iterative proportional scaling via decomposable submodels for contingency tables
• Unified generalized iterative scaling and its applications
• Directionally collapsible parameterizations of multivariate binary distributions
• Classical iterative proportional scaling of log-linear models with rational maximum likelihood estimator
• Estimation and testing of multiplicative models for frequency data
• scientific article; zbMATH DE number 4100445 (Why is no real title available?)
• On the role of the overall effect in exponential families
• On the closure of relational models
• Hierarchical Aitchison-Silvey models for incomplete binary sample spaces
• Testing the fit of relational models
• A fast algorithm for iterative proportional fitting in log-linear models
• Iterative Proportional Scaling Revisited: A Modern Optimization Perspective

Describes a project that uses

Uses Software

• gIPFrm
• R

This page was built for publication: Iterative scaling in curved exponential families

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2949880)