Recursive bias estimation for high dimensional smoothers

Journal Article · Tue Jan 01 04:00:00 EST 2008 · Annuals of Statistics

OSTI ID:960889

Hengartner, Nicolas W ^[1]; Matzner-lober, Eric ^[2]; Cornillon, Pierre - Andre ^[3]

---

1. Los Alamos National Laboratory
2. UHB, FRANCE
3. INRA

In multivariate nonparametric analysis, sparseness of the covariates also called curse of dimensionality, forces one to use large smoothing parameters. This leads to biased smoothers. Instead of focusing on optimally selecting the smoothing parameter, we fix it to some reasonably large value to ensure an over-smoothing of the data. The resulting smoother has a small variance but a substantial bias. In this paper, we propose to iteratively correct the bias initial estimator by an estimate of the latter obtained by smoothing the residuals. We examine in detail the convergence of the iterated procedure for classical smoothers and relate our procedure to L{sub 2}-Boosting. We apply our method to simulated and real data and show that our method compares favorably with existing procedures.

• Research Organization: Los Alamos National Laboratory (LANL)
• Sponsoring Organization: DOE
• DOE Contract Number: AC52-06NA25396
• OSTI ID: 960889
• Report Number(s): LA-UR-08-06365; LA-UR-08-6365
• Journal Information: Annuals of Statistics, Journal Name: Annuals of Statistics
• Country of Publication: United States
• Language: English

Related Subjects

97 MATHEMATICS AND COMPUTING
CONVERGENCE
DATA
DATA ANALYSIS
FOCUSING
REGRESSION ANALYSIS