On Mean Estimation for Heteroscedastic Random Variables

DOI10.1214/21-AIHP1239MaRDI QIDQ6351968arXivFDO

Authors Luc Devroye, Silvio Lattanzi, Gábor Lugosi, Nikita Zhivotovskiy

Publication date 22 October 2020

Mathematics Subject Classification ID

Parametric tolerance and confidence regions (62F25) Robustness and adaptive procedures (parametric inference) (62F35) Order statistics; empirical distribution functions (62G30)

Abstract: We study the problem of estimating the common mean $mu$ of $n$ independent symmetric random variables with different and unknown standard deviations $sigm{a}_{1}lesigm{a}_{2}lecdotslesigm{a}_n$. We show that, under some mild regularity assumptions on the distribution, there is a fully adaptive estimator $widehatmu$ such that it is invariant to permutations of the elements of the sample and satisfies that, up to logarithmic factors, with high probability, [ |widehat{mu} - mu| lesssim minleft{sigma_{m^*}, frac{sqrt{n}}{sum_{i = sqrt{n}}^n sigma_i^{-1}} ight}~, ] where the index $m^{\ast }lesssimsqrtn$ satisfies $m^{\ast }approxsqrtsigm{a}_{m^{\ast }}su{m}_{i=m^{\ast }}^{n}sigm{a}_i^{-1}$.

This page was built for publication: On Mean Estimation for Heteroscedastic Random Variables

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6351968)