52
8

Some new ideas in nonparametric estimation

Abstract

In the framework of an abstract statistical model we discuss how to use the solution of one estimation problem ({\it Problem A}) in order to construct an estimator in another, completely different, {\it Problem B}. As a solution of {\it Problem A} we understand a data-driven selection from a given family of estimators A(\mH)={A^\mh,\mh\mH}\mathbf{A}(\mH)=\big\{\widehat{A}_\mh, \mh\in\mH\big\} and establishing for the selected estimator so-called oracle inequality. %parameterized by some se t\mH\mH. If \mh^\mH\hat{\mh}\in\mH is the selected parameter and B(\mH)={B^\mh,\mh\mH}\mathbf{B}(\mH)=\big\{\widehat{B}_\mh, \mh\in\mH\big\} is an estimator's collection built in {\it Problem B} we suggest to use the estimator B^\mh^\widehat{B}_{\hat{\mh}}. We present very general selection rule led to selector \mh^\hat{\mh} and find conditions under which the estimator B^\mh^\widehat{B}_{\hat{\mh}} is reasonable. Our approach is illustrated by several examples related to adaptive estimation.

View on arXiv
Comments on this paper