Consistency and rates of convergence of k/sub n/ nearest neighbor estimation under arbitrary sampling

Abstract

Consistency and rates of convergence of the k/sub n/-NN estimator are established in the general case in which samples are chosen arbitrarily from a compact metric space.