Wild bootstrap bandwidth selection of recursive nonparametric relative regression for independent functional data

We propose and investigate a new kernel regression estimator based on the minimization of the mean squared relative error. We study the properties of the proposed recursive estimator and compare it with the recursive estimator based on the minimization of the mean squared error proposed by Slaoui (2018). It turns out that, with an adequate choice of the parameters, the proposed estimator performs better than the recursive estimator based on the minimization of the mean squared error. We illustrate these theoretical results through a real chemometric dataset. (C) 2019 Elsevier Inc. All rights reserved.