Cross-validation and non-parametric k nearest-neighbour estimation
Academic Article
Overview
Research
Identity
Additional Document Info
Other
View All
Overview
abstract
In this paper we consider the problem of estimating a non-parametric regression function using the k nearest-neighbour method. We provide asymptotic theories for the least-squares cross validation (CV) selected smoothing parameter k for both local constant and local linear estimation methods. We also establish the asymptotic normality results for the resulting non-parametric regression function estimators. Some limited Monte Carlo experiments show that the CV method performs well in finite sample applications. Royal Economic Society 2006.