English

PCA-Kernel Estimation

Statistics Theory 2010-04-26 v1 Statistics Theory

Abstract

Many statistical estimation techniques for high-dimensional or functional data are based on a preliminary dimension reduction step, which consists in projecting the sample \bX1,\hdots,\bXn\bX_1, \hdots, \bX_n onto the first DD eigenvectors of the Principal Component Analysis (PCA) associated with the empirical projector Π^D\hat \Pi_D. Classical nonparametric inference methods such as kernel density estimation or kernel regression analysis are then performed in the (usually small) DD-dimensional space. However, the mathematical analysis of this data-driven dimension reduction scheme raises technical problems, due to the fact that the random variables of the projected sample (Π^D\bX1,\hdots,Π^D\bXn)(\hat \Pi_D\bX_1,\hdots, \hat \Pi_D\bX_n) are no more independent. As a reference for further studies, we offer in this paper several results showing the asymptotic equivalencies between important kernel-related quantities based on the empirical projector and its theoretical counterpart. As an illustration, we provide an in-depth analysis of the nonparametric kernel regression case

Keywords

Cite

@article{arxiv.1003.5089,
  title  = {PCA-Kernel Estimation},
  author = {Gérard Biau and André Mas},
  journal= {arXiv preprint arXiv:1003.5089},
  year   = {2010}
}
R2 v1 2026-06-21T15:02:57.844Z