Track: Statistics and Empirical Research
Abstract
This paper establishes exponential inequalities for the probability of the distance between kernel estimator and its means in nonparametric regression problem with mixing variables. We consider an operator equation taking the following form Y=Aq(Z)+e, where A is a compact operator.
The goal is to estimate the functional q when the variable Z is contaminated by measurements errors.
Keywords: convolution linear compact operator; kernel estimator; mixing process; non parametric regression.