(A) method for biased estimate of bandwidth in nonparametric regression

Abstract

This paper is concerned with the problem of choosing a bandwidth for nonparametric regression. We consider a method based on an biased estimate of mean average squared error. We show that the bandwidth chosen our method is asymptotically optimal and has small sample variability. In simulation study, we show that our method is very closer to optimum bandwidth than any other method when the underlying regression function is sufficiently smooth.