Functional additive expectile regression in the reproducing kernel Hilbert space

Abstract

In the literature, the functional additive regression model has received much attention. Most current studies, however, only estimate the mean function, which may not adequately capture the heteroscedasticity and/or asymmetries of the model errors. In light of this, we extend functional additive regression models to their expectile counterparts and obtain an upper bound on the convergence rate of its regularized estimator under mild conditions. To demonstrate its finite sample performance, a few simulation experiments and a real data example are provided.