Consider the heteroscedastic model Y=m(X)+[sigma](X)[var epsilon], where [var epsilon] and X are independent, Y is subject to right censoring, m(·) is an unknown but smooth location function (like e.g. conditional mean, median, trimmed mean...) and [sigma](·) an unknown but smooth scale...

Suppose the random vector (X,Y) satisfies the regression model Y=m(X)+[sigma](X)[epsilon], where m([dot operator]) is the conditional mean, [sigma]2([dot operator]) is the conditional variance, and [epsilon] is independent of X. The covariate X is d-dimensional (d=1), the response Y is...

Assume that we have two populations (X <Subscript>1</Subscript>,Y <Subscript>1</Subscript>) and (X <Subscript>2</Subscript>,Y <Subscript>2</Subscript>) satisfying two general nonparametric regression models Y <Subscript> j </Subscript>=m <Subscript> j </Subscript>(X <Subscript> j </Subscript>)+ε <Subscript> j </Subscript>, j=1,2, where m(⋅) is a smooth location function, ε <Subscript> j </Subscript> has zero location and the response Y <Subscript> j </Subscript> is possibly right-censored. In this paper, we propose...</subscript></subscript></subscript></subscript></subscript></subscript></subscript></subscript></subscript></subscript>