Séminaires

Strong uniform consistency of the local linear relative error regression estimator under left truncation

par Elias Ould Saïd (IUT Calais)

Europe/Paris
GI - UTC

GI - UTC

Description

In this talk, we will discuss about  a nonparametric estimator of the regression function based 

on the local linear method when the loss function is the mean squared relative error and the data left truncated. 

The proposed method avoids the problem of boundary effects and reduces the bias term in addition to being 

robust against the presence of outliers. Under suitable assumptions, the estimator’s strong uniform 

almost sure consistency with rate is established and its finite-sample properties are investigated 

by means of a simulation study.