One-sided cross-validation for nonsmooth regression functions

Olga Y. Savchuk; Jeffrey D. Hart; Simon P. Sheather

doi:10.1080/10485252.2013.817575

One-sided cross-validation for nonsmooth regression functions

Olga Y. Savchuk, Jeffrey D. Hart, Simon P. Sheather

Research output: Contribution to journal › Article › peer-review

2 Scopus citations

Abstract

The one-sided cross-validation (OSCV) method is shown to be robust to lack of smoothness in the regression function. Two corrections for the case where the regression function has a discontinuous first derivative are proposed. Simulation results suggest that proposed modifications of the OSCV method are efficient for regression functions whose first derivative is discontinuous at more than two points. The OSCV method and its modification outperform the cross-validation method and the Ruppert-Sheather-Wand plug-in method in a data example involving a function that, potentially, has one discontinuity in its derivative.

Original language	English
Pages (from-to)	889-904
Number of pages	16
Journal	Journal of Nonparametric Statistics
Volume	25
Issue number	4
DOIs	https://doi.org/10.1080/10485252.2013.817575
State	Published - Dec 2013

Bibliographical note

Funding Information:
The authors thank two anonymous referees whose comments led to the substantial improvement of the article. The work of Professor Hart was supported by NSF grant DMS 1106694.

Funding

The authors thank two anonymous referees whose comments led to the substantial improvement of the article. The work of Professor Hart was supported by NSF grant DMS 1106694.

Funders	Funder number
National Science Foundation (NSF)	DMS 1106694

Keywords

average squared error
bandwidth selection
cross-validation
local linear estimator
mean average squared error

ASJC Scopus subject areas

Statistics and Probability
Statistics, Probability and Uncertainty

Access to Document

10.1080/10485252.2013.817575

Cite this

@article{3921c9dcc45a4e6b92524bbdb06b6da1,

title = "One-sided cross-validation for nonsmooth regression functions",

abstract = "The one-sided cross-validation (OSCV) method is shown to be robust to lack of smoothness in the regression function. Two corrections for the case where the regression function has a discontinuous first derivative are proposed. Simulation results suggest that proposed modifications of the OSCV method are efficient for regression functions whose first derivative is discontinuous at more than two points. The OSCV method and its modification outperform the cross-validation method and the Ruppert-Sheather-Wand plug-in method in a data example involving a function that, potentially, has one discontinuity in its derivative.",

keywords = "average squared error, bandwidth selection, cross-validation, local linear estimator, mean average squared error",

author = "Savchuk, {Olga Y.} and Hart, {Jeffrey D.} and Sheather, {Simon P.}",

note = "Funding Information: The authors thank two anonymous referees whose comments led to the substantial improvement of the article. The work of Professor Hart was supported by NSF grant DMS 1106694.",

year = "2013",

month = dec,

doi = "10.1080/10485252.2013.817575",

language = "English",

volume = "25",

pages = "889--904",

number = "4",

}

TY - JOUR

T1 - One-sided cross-validation for nonsmooth regression functions

AU - Savchuk, Olga Y.

AU - Hart, Jeffrey D.

AU - Sheather, Simon P.

N1 - Funding Information: The authors thank two anonymous referees whose comments led to the substantial improvement of the article. The work of Professor Hart was supported by NSF grant DMS 1106694.

PY - 2013/12

Y1 - 2013/12

N2 - The one-sided cross-validation (OSCV) method is shown to be robust to lack of smoothness in the regression function. Two corrections for the case where the regression function has a discontinuous first derivative are proposed. Simulation results suggest that proposed modifications of the OSCV method are efficient for regression functions whose first derivative is discontinuous at more than two points. The OSCV method and its modification outperform the cross-validation method and the Ruppert-Sheather-Wand plug-in method in a data example involving a function that, potentially, has one discontinuity in its derivative.

AB - The one-sided cross-validation (OSCV) method is shown to be robust to lack of smoothness in the regression function. Two corrections for the case where the regression function has a discontinuous first derivative are proposed. Simulation results suggest that proposed modifications of the OSCV method are efficient for regression functions whose first derivative is discontinuous at more than two points. The OSCV method and its modification outperform the cross-validation method and the Ruppert-Sheather-Wand plug-in method in a data example involving a function that, potentially, has one discontinuity in its derivative.

KW - average squared error

KW - bandwidth selection

KW - cross-validation

KW - local linear estimator

KW - mean average squared error

UR - http://www.scopus.com/inward/record.url?scp=84885383715&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84885383715&partnerID=8YFLogxK

U2 - 10.1080/10485252.2013.817575

DO - 10.1080/10485252.2013.817575

M3 - Article

AN - SCOPUS:84885383715

SN - 1048-5252

VL - 25

SP - 889

EP - 904

JO - Journal of Nonparametric Statistics

JF - Journal of Nonparametric Statistics

IS - 4

ER -

One-sided cross-validation for nonsmooth regression functions

Abstract

Bibliographical note

Funding

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this