In the panoply of pattern classification techniques, few enjoy the intuitive
appeal and simplicity of the nearest neighbor rule: given a set of samples in
some metric domain space whose value under some function is known, we estimate
the function anywhere in the domain by giving the va