Robustness of one-sided cross-validation to autocorrelation Academic Article

abstract

• The effects of moderate levels of serial correlation on one-sided and ordinary cross-validation in the context of local linear and kernel smoothing is investigated. It is shown both theoretically and by simulation that one-sided cross-validation is much less adversely affected by correlation than is ordinary cross-validation. The former method is a reliable means of window width selection in the presence of moderate levels of serial correlation, while the latter is not. It is also shown that ordinary cross-validation is less robust to correlation when applied to Gasser-Mller kernel estimators than to local linear ones. 2003 Elsevier Inc. All rights reserved.

authors

• Hart, Jeffrey

published proceedings

• JOURNAL OF MULTIVARIATE ANALYSIS

author list (cited authors)

• Hart, J. D., & Lee, C. L.

citation count

• 17

complete list of authors

• Hart, JD||Lee, CL

publication date

• January 2005

publisher

• Elsevier  Publisher

published in

• Journal of Multivariate Analysis  Journal

keywords

• AUTOREGRESSIVE PROCESS
• Average Squared Error
• Data-driven Smoothing Parameters
• Nonparametric Regression

Digital Object Identifier (DOI)

• 10.1016/j.jmva.2003.08.005

start page

• 77

end page

• 96

volume

• 92

issue

• 1

URL

• http%3A%2F%2Fdx.doi.org%2F10.1016%2Fj.jmva.2003.08.005