Multivariate boundary regression models

In this work, we consider a multivariate regression model with one-sided errors. We assume for the function to lie in general Hölder class and estimate it via nonparametric local polynomial approach that consists of minimization integral approximation lying above data points. While consideration covariates offers an undeniable opportunity from application-oriented standpoint, requires new method proof replace established ones univariate case. The main purpose paper is show uniform consistency provide rates convergence considered estimator both random deterministic design To demonstrate performance estimators, small sample behavior investigated simulation study dimension two three.