Using local linear kernel smoothers to test the lack of fit of nonlinear regression models

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

Herein, we propose a data-driven test that assesses the lack of fit of nonlinear regression models. The comparison of local linear kernel and parametric fits is the basis of this test, and specific boundary-corrected kernels are not needed at the boundary when local linear fitting is used. Under the parametric null model, the asymptotically optimal bandwidth can be used for bandwidth selection. This selection method leads to the data-driven test that has a limiting normal distribution under the null hypothesis and is consistent against any fixed alternative. The finite-sample property of the proposed data-driven test is illustrated, and the power of the test is compared with that of some existing tests via simulation studies. We illustrate the practicality of the proposed test by using two data sets.

Original languageEnglish (US)
Pages (from-to)267-284
Number of pages18
JournalStatistical Methodology
Volume2
Issue number4
DOIs
StatePublished - Dec 2005
Externally publishedYes

Fingerprint

Lack of Fit
Nonlinear Regression Model
kernel
Data-driven
Optimal Bandwidth
Bandwidth Selection
Asymptotically Optimal
Limiting Distribution
Null hypothesis
Null
Gaussian distribution
Simulation Study
Alternatives

Keywords

  • Bandwidth selection
  • Boundary effects
  • Fit comparison
  • Local linear kernel estimator

ASJC Scopus subject areas

  • Statistics and Probability

Cite this

Using local linear kernel smoothers to test the lack of fit of nonlinear regression models. / Li, Chin-Shang.

In: Statistical Methodology, Vol. 2, No. 4, 12.2005, p. 267-284.

Research output: Contribution to journalArticle

@article{f63484651d234bc1ae02581e96004cc7,
title = "Using local linear kernel smoothers to test the lack of fit of nonlinear regression models",
abstract = "Herein, we propose a data-driven test that assesses the lack of fit of nonlinear regression models. The comparison of local linear kernel and parametric fits is the basis of this test, and specific boundary-corrected kernels are not needed at the boundary when local linear fitting is used. Under the parametric null model, the asymptotically optimal bandwidth can be used for bandwidth selection. This selection method leads to the data-driven test that has a limiting normal distribution under the null hypothesis and is consistent against any fixed alternative. The finite-sample property of the proposed data-driven test is illustrated, and the power of the test is compared with that of some existing tests via simulation studies. We illustrate the practicality of the proposed test by using two data sets.",
keywords = "Bandwidth selection, Boundary effects, Fit comparison, Local linear kernel estimator",
author = "Chin-Shang Li",
year = "2005",
month = "12",
doi = "10.1016/j.stamet.2005.06.001",
language = "English (US)",
volume = "2",
pages = "267--284",
journal = "Statistical Methodology",
issn = "1572-3127",
publisher = "Elsevier",
number = "4",

}

TY - JOUR

T1 - Using local linear kernel smoothers to test the lack of fit of nonlinear regression models

AU - Li, Chin-Shang

PY - 2005/12

Y1 - 2005/12

N2 - Herein, we propose a data-driven test that assesses the lack of fit of nonlinear regression models. The comparison of local linear kernel and parametric fits is the basis of this test, and specific boundary-corrected kernels are not needed at the boundary when local linear fitting is used. Under the parametric null model, the asymptotically optimal bandwidth can be used for bandwidth selection. This selection method leads to the data-driven test that has a limiting normal distribution under the null hypothesis and is consistent against any fixed alternative. The finite-sample property of the proposed data-driven test is illustrated, and the power of the test is compared with that of some existing tests via simulation studies. We illustrate the practicality of the proposed test by using two data sets.

AB - Herein, we propose a data-driven test that assesses the lack of fit of nonlinear regression models. The comparison of local linear kernel and parametric fits is the basis of this test, and specific boundary-corrected kernels are not needed at the boundary when local linear fitting is used. Under the parametric null model, the asymptotically optimal bandwidth can be used for bandwidth selection. This selection method leads to the data-driven test that has a limiting normal distribution under the null hypothesis and is consistent against any fixed alternative. The finite-sample property of the proposed data-driven test is illustrated, and the power of the test is compared with that of some existing tests via simulation studies. We illustrate the practicality of the proposed test by using two data sets.

KW - Bandwidth selection

KW - Boundary effects

KW - Fit comparison

KW - Local linear kernel estimator

UR - http://www.scopus.com/inward/record.url?scp=33644773761&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33644773761&partnerID=8YFLogxK

U2 - 10.1016/j.stamet.2005.06.001

DO - 10.1016/j.stamet.2005.06.001

M3 - Article

AN - SCOPUS:33644773761

VL - 2

SP - 267

EP - 284

JO - Statistical Methodology

JF - Statistical Methodology

SN - 1572-3127

IS - 4

ER -