Linear Information Models: An Introduction

10.6339/JDS.2007.05(3).442

Philip E. Cheng；Jiun W. Liou；Michelle Liou；John A. D. Aston

Contingency tables ； log-linear models ； information models ； model selection ； mutual information

Journal of Data Science

5卷3期（2007 / 07 / 01）

297 - 313

英文

Relative entropy identities yield basic decompositions of categorical data log-likelihoods. These naturally lead to the development of information models in contrast to the hierarchical log-linear models. A recent study by the authors clarified the principal difference in the data likelihood analysis between the two model types. The proposed scheme of log-likelihood decomposition introduces a prototype of linear information models, with which a basic scheme of model selection can be formulated accordingly. Empirical studies with high-way contingency tables are exemplified to illustrate the natural selections of information models in contrast to hierarchical log-linear models.