We wish to make inferences about the conditional probabilities $p(y|x)$, many of which are zero, when the distribution of X is unknown and one observes only a multinomial sample of the Y variates. To do this, fixed likelihood ratio models and quasi-incremental distributions are defined. It is shown that quasi-incremental distributions are intimately linked to decomposable graphs and that these graphs can guide us to transformations of X and Y which admit a conjugate Bayesian analysis on a reparametrization of the conditional probabilities of interest.
Publié le : 1996-10-14
Classification:
Bayesian probability estimation,
constraint graph,
contingency tables,
decomposable graph,
generalized Dirichlet distributions,
separation of likelihood,
62F15,
62H17
@article{1069362316,
author = {Smith, Jim Q. and Queen, Catriona M.},
title = {Bayesian models for sparse probability tables},
journal = {Ann. Statist.},
volume = {24},
number = {6},
year = {1996},
pages = { 2178-2198},
language = {en},
url = {http://dml.mathdoc.fr/item/1069362316}
}
Smith, Jim Q.; Queen, Catriona M. Bayesian models for sparse probability tables. Ann. Statist., Tome 24 (1996) no. 6, pp. 2178-2198. http://gdmltest.u-ga.fr/item/1069362316/