Abstract
Fully robust OSCV is a modification of the OSCV method that produces consistent bandwidths in the cases of smooth and nonsmooth regression functions. We propose the practical implementation of the method based on the robust cross-validation kernel \(H_I\) in the case when the Gaussian kernel \(\phi \) is used in computing the resulting regression estimate. The kernel \(H_I\) produces practically unbiased bandwidths in the smooth and nonsmooth cases and performs adequately in the data examples. Negative tails of \(H_I\) occasionally result in unacceptably wiggly OSCV curves in the neighborhood of zero. This problem can be resolved by selecting the bandwidth from the largest local minimum of the curve. Further search for the robust kernels with desired properties brought us to consider the quartic kernel for the cross-validation purposes. The quartic kernel is almost robust in the sense that in the nonsmooth case it substantially reduces the asymptotic relative bandwidth bias compared to \(\phi \). However, the quartic kernel is found to produce more variable bandwidths compared to \(\phi \). Nevertheless, the quartic kernel has an advantage of producing smoother OSCV curves compared to \(H_I\). A simplified scale-free version of the OSCV method based on a rescaled one-sided kernel is proposed.
Similar content being viewed by others
References
Cleveland WS (1979) Robust locally weighted regression and smoothing scatterplots. J Am Stat Assoc 74(368):829–836 (ISSN 0003-1291)
Dudzinski M, Mykytowycz R (1961) The eye lens as an indicator of age in the wild rabbit in Australia. CSIRO Wildl Res 6:156–159
Gasser T, Müller H-G (1979) Kernel estimation of regression functions. In: Smoothing techniques for curve estimation, Proceedings Workshop, Heidelberg, volume 757 of Lecture Notes in Mathematics. Springer, Berlin, pp 23–68
Hall P, Marron JS (1991) Local minima in cross-validation functions. J R Stat Soc Ser B 53(1):245–252
Hart JD, Lee C-L (2005) Robustness of one-sided cross-validation to autocorrelation. J Multivar Anal 92(1):77–96 (ISSN 0047-259X)
Hart JD, Yi S (1998) One-sided cross-validation. J Am Stat Assoc 93(442):620–631
Köhler M, Schindler A, Sperlich S (2014) A review and comparison of bandwidth selection methods for kernel regression. Int Stat Rev 82(2):243–274
Lichman M (2013) UCI machine learning repository. http://archive.ics.uci.edu/ml
Mammen E, Martínez Miranda MD, Nielsen JP, Sperlich S (2011) Do-validation for kernel density estimation. J Am Stat Assoc 106(494):651–660. doi:10.1198/jasa.2011.tm08687 (ISSN 0162-1459)
Mammen E, Martínez Miranda MD, Nielsen JP, Sperlich S (2014) Further theoretical and practical insight to the do-validated bandwidth selector. J Korean Stat Soc 43(3):355–365
Martínez-Miranda MD, Nielsen JP, Sperlich S (2009) One sided cross validation for density estimation. In: Gregoriou GN (ed) Operational risk towards basel III: best practices and issues in modeling, management and regulation. Wiley, Hoboken, pp 177–196
Ruppert D, Sheather SJ, Wand MP (1995) An effective bandwidth selector for local least squares regression. J Am Stat Assoc 90(432):1257–1270 (ISSN 0162-1459)
Savchuk OY, Hart JD, Sheather SJ (2010) Indirect cross-validation for density estimation. J Am Stat Assoc 105(489):415–423
Savchuk OY, Hart JD, Sheather SJ (2013) One-sided cross-validation for nonsmooth regression functions. J Nonparametr Stat 25(4):889–904
Savchuk OY, Hart JD, Sheather SJ (2016) Corrigendum to “One-sided cross-validation for nonsmooth regression functions” [J Nonparametr Stat 25(4):889–904 (2013)]. J Nonparametr Stat 28(4):875–877
Stone CJ (1977) Consistent nonparametric regression. Ann Stat 5(4):595–645 (ISSN 0090-5364. With discussion and a reply by the author)
Wand MP, Jones MC (1995) Kernel smoothing, volume 60 of Monographs on statistics and applied probability. Chapman and Hall Ltd., London
Weisberg S (2005) Applied linear regression, 3rd edn. Wiley series in probability and statistics. Wiley-Interscience, Hoboken. doi:10.1002/0471704091 (ISBN 0-471-66379-4)
Yi S (1996) On one-sided cross-validation in nonparametric regression. Ph.D dissertation, Texas A&M University
Yi S (2005) A comparison of two bandwidth selectors OSCV and AICc in nonparametric regression. Commun Stat Simul Comput 34(3):585–594 (ISSN 0361-0918)
Author information
Authors and Affiliations
Corresponding author
Appendix
Appendix
Notation For an arbitrary function g, define the following functionals:
where for all z,
Regression functions Regression functions \(r_1\), \(r_2\), and \(r_3\) are defined below. For each function, \(0\le x\le 1\).
Rights and permissions
About this article
Cite this article
Savchuk, O.Y., Hart, J.D. Fully robust one-sided cross-validation for regression functions. Comput Stat 32, 1003–1025 (2017). https://doi.org/10.1007/s00180-017-0713-7
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00180-017-0713-7