Consistency of Bayes estimates for nonparametric regression: normal theory
Diaconis, Persi W. ; Freedman, David
Bernoulli, Tome 4 (1998) no. 1, p. 411-444 / Harvested from Project Euclid
Performance characteristics of Bayes estimates are studied. More exactly, for each subject in a data set, let ξ be a vector of binary covariates and let Y be a normal response variable, with E{Y|ξ}=f(ξ) and var{Y|ξ}=1. Here, f is an unknown function to be estimated from the data; the subjects are independent and identically distributed. Define a prior distribution on f as ∑kwkπk/∑kwk, where πk is standard normal on the set of f which only depend on the first k covariates and wk>0 for infinitely many k. Bayes estimates are consistent for all f. On the other hand, if the πk are flat, inconsistency is the rule.
Publié le : 1998-12-14
Classification:  Bayes estimates,  binary regression,  consistency,  model selection
@article{1173883814,
     author = {Diaconis, Persi W. and Freedman, David},
     title = {Consistency of Bayes estimates for nonparametric regression: normal theory},
     journal = {Bernoulli},
     volume = {4},
     number = {1},
     year = {1998},
     pages = { 411-444},
     language = {en},
     url = {http://dml.mathdoc.fr/item/1173883814}
}
Diaconis, Persi W.; Freedman, David. Consistency of Bayes estimates for nonparametric regression: normal theory. Bernoulli, Tome 4 (1998) no. 1, pp.  411-444. http://gdmltest.u-ga.fr/item/1173883814/