Strong uniform consistency of nonparametric regression function estimates (Q1111287): Difference between revisions

Let (X,Y) be an \({\mathbb{R}}^ d\times {\mathbb{R}}\)-valued random vector and let \(r(t)=E(Y| X=t)\) be the regression function of Y on X that has to be estimated from a sample \((X_ i,Y_ i)\), \(i=1,...,n\). We establish conditions ensuring that an estimate of the form \[ r_ n(t)=\sum^{n}_{i=1}Y_ i\phi_{ni}(t,X_ i)/\sum^{n}_{i=1}\phi_{ni}(t,X_ i), \] where \(\phi_{ni}(t,x)\) is a sequence of Borel measurable functions on \({\mathbb{R}}^ d\times {\mathbb{R}}^ d\), is uniformly strongly consistent with a certain rate of convergence. By applying this result we obtain rates of strong uniform consistency of the regressogram, kernel estimates, \(k_ n\)-nearest neighbor estimates and estimates based on orthogonal series.