Geodesic
Iterative isotonic regression
Guyader, Arnaud1  ; Hengartner, Nick2 ; Jégou, Nicolas3  ; Matzner-Løber, Eric3 
1 UniversitéRennes 2, INRIA and IRMAR, Campus de Villejean, 35043 Rennes, France
2 Los Alamos National Laboratory, NM 87545, Los Alamos, USA
3 Université Rennes 2, Campus de Villejean, 35043 Rennes, France
ESAIM: Probability and Statistics, Tome 19 (2015), pp. 1-23

Voir la notice de l'article provenant de la source Numdam

This article explores some theoretical aspects of a recent nonparametric method for estimating a univariate regression function of bounded variation. The method exploits the Jordan decomposition which states that a function of bounded variation can be decomposed as the sum of a non-decreasing function and a non-increasing function. This suggests combining the backfitting algorithm for estimating additive functions with isotonic regression for estimating monotone functions. The resulting iterative algorithm is called Iterative Isotonic Regression (I.I.R.). The main result in this paper states that the estimator is consistent if the number of iterations k n grows appropriately with the sample size n. The proof requires two auxiliary results that are of interest in and by themselves: firstly, we generalize the well-known consistency property of isotonic regression to the framework of a non-monotone regression function, and secondly, we relate the backfitting algorithm to von Neumann’s algorithm in convex analysis. We also analyse how the algorithm can be stopped in practice using a data-splitting procedure.

Reçu le :
DOI : 10.1051/ps/2014012
Classification : 52A05, 62G08, 62G20
Keywords: Nonparametric statistics, isotonic regression, additive models, metric projection onto convex cones

Guyader, Arnaud 1 ; Hengartner, Nick 2 ; Jégou, Nicolas 3 ; Matzner-Løber, Eric 3

1 UniversitéRennes 2, INRIA and IRMAR, Campus de Villejean, 35043 Rennes, France
2 Los Alamos National Laboratory, NM 87545, Los Alamos, USA
3 Université Rennes 2, Campus de Villejean, 35043 Rennes, France 
@article{PS_2015__19__1_0,
     author = {Guyader, Arnaud and Hengartner, Nick and J\'egou, Nicolas and Matzner-L{\o}ber, Eric},
     title = {Iterative isotonic regression},
     journal = {ESAIM: Probability and Statistics},
     pages = {1--23},
     publisher = {EDP-Sciences},
     volume = {19},
     year = {2015},
     doi = {10.1051/ps/2014012},
     mrnumber = {3374866},
     zbl = {1392.62117},
     language = {en},
     url = {http://geodesic.mathdoc.fr/articles/10.1051/ps/2014012/}
}
TY  - JOUR
AU  - Guyader, Arnaud
AU  - Hengartner, Nick
AU  - Jégou, Nicolas
AU  - Matzner-Løber, Eric
TI  - Iterative isotonic regression
JO  - ESAIM: Probability and Statistics
PY  - 2015
SP  - 1
EP  - 23
VL  - 19
PB  - EDP-Sciences
UR  - http://geodesic.mathdoc.fr/articles/10.1051/ps/2014012/
DO  - 10.1051/ps/2014012
LA  - en
ID  - PS_2015__19__1_0
ER  - 
%0 Journal Article
%A Guyader, Arnaud
%A Hengartner, Nick
%A Jégou, Nicolas
%A Matzner-Løber, Eric
%T Iterative isotonic regression
%J ESAIM: Probability and Statistics
%D 2015
%P 1-23
%V 19
%I EDP-Sciences
%U http://geodesic.mathdoc.fr/articles/10.1051/ps/2014012/
%R 10.1051/ps/2014012
%G en
%F PS_2015__19__1_0
Guyader, Arnaud; Hengartner, Nick; Jégou, Nicolas; Matzner-Løber, Eric. Iterative isotonic regression. ESAIM: Probability and Statistics, Tome 19 (2015), pp. 1-23. doi: 10.1051/ps/2014012

Cité par Sources :