English

Nonparametric Estimation of the Regression Function in an Errors-in-Variables Model

Statistics Theory 2008-02-11 v1 Statistics Theory

Abstract

We consider the regression model with errors-in-variables where we observe nn i.i.d. copies of (Y,Z)(Y,Z) satisfying Y=f(X)+ξ,Z=X+σϵY=f(X)+\xi, Z=X+\sigma\epsilon, involving independent and unobserved random variables X,ξ,ϵX,\xi,\epsilon. The density gg of XX is unknown, whereas the density of σϵ\sigma\epsilon is completely known. Using the observations (Y_i,Z_i)(Y\_i, Z\_i), i=1,...,ni=1,...,n, we propose an estimator of the regression function ff, built as the ratio of two penalized minimum contrast estimators of =fg\ell=fg and gg, without any prior knowledge on their smoothness. We prove that its L_2\mathbb{L}\_2-risk on a compact set is bounded by the sum of the two L_2(R)\mathbb{L}\_2(\mathbb{R})-risks of the estimators of \ell and gg, and give the rate of convergence of such estimators for various smoothness classes for \ell and gg, when the errors ϵ\epsilon are either ordinary smooth or super smooth. The resulting rate is optimal in a minimax sense in all cases where lower bounds are available.

Keywords

Cite

@article{arxiv.math/0511111,
  title  = {Nonparametric Estimation of the Regression Function in an Errors-in-Variables Model},
  author = {Fabienne Comte and Marie-Luce Taupin},
  journal= {arXiv preprint arXiv:math/0511111},
  year   = {2008}
}