Nonparametric quantile regression for spatio-temporal processes
Authors: Soudeep Deb, Claudia Neves, Subhrajyoty Roy
Summary: On this paper, we develop a brand new and efficient method to nonparametric quantile regression that accommodates ultrahigh-dimensional information arising from spatio-temporal processes. This method proves advantageous in staving off computational challenges that represent recognized hindrances to present nonparametric quantile regression strategies when the variety of predictors is far bigger than the accessible pattern measurement. We examine circumstances underneath which estimation is possible and of fine general high quality and procure sharp approximations that we make use of to devising statistical inference methodology. These embody simultaneous confidence intervals and exams of hypotheses, whose asymptotics is borne by a non-trivial practical central restrict theorem tailor-made to martingale variations. Moreover, we offer finite-sample outcomes by means of varied simulations which, accompanied by an illustrative utility to real-worldesque information (on electrical energy demand), provide ensures on the efficiency of the proposed methodology