The Estimation of Residual Variance in Nonparametric Regression
DOI:
https://doi.org/10.20956/j.v17i3.13192Keywords:
Nonparametric regression, Rice estimator, GSJ estimator, Tong-Wang estimatorAbstract
Given a nonparametric regression model
Yi = g(xi) + ei, i = 1, 2, …, n,
where Y is a dependent variable, x is an independent variable, g is an unknown function and e is an error assumed to be an independent, identical, and is distributed with mean 0 and variance σ2. In this research Rice estimator is used to determine the biased value of a residual variance estimator. The Rice estimator is given as follows:
.
The biased value of residual variance estimator of the Rice method is:
, where and.
Using the Rice estimator, the Tong-Wang residual variance estimator is obtained, that is:
,
Where , , ,
, , k = 1, 2, … , m.
Based upon the data simulation by considering the exponential, arithmetical, and trigonometrical models, it is found that the MSE value of the Tong-Wang estimator tends to be less compared to those of the Rice estimator as well as the GSJ (Gasser, Sroka, and Jennen) estimator.
References
Buckley, M. J., Eagleson, G. K., & Silverman, B. W. 1988. The Estimation of Residual Variance in Nonparametric Regression. Biometrika, 75(2), 189. https://doi.org/10.2307/2336166
Dette, H., Munk, A., & Wagner, T. 1998. Estimating the variance in nonparametric regression - what is a reasonable choice? Journal of the Royal Statistical Society B, 60(4), 751–764.
Drygas, H. 1972. The estimation of residual variance in regression analysis. Mathematische Operationsforschung Und Statistik, 3(5), 373–388. https://doi.org/10.1080/02331887208801094
Evans, D., & Jones, A. J. 2008. Non-parametric estimation of residual moments and covariance. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 464(2099), 2831–2846. https://doi.org/10.1098/rspa.2007.0195
GASSER, T., SROKA, L., & JENNEN-STEINMETZ, C. 1986. Residual variance and residual pattern in nonlinear regression. Biometrika, 73(3), 625–633. https://doi.org/10.1093/biomet/73.3.625
Liitiäinen, E., Verleysen, M., Corona, F., & Lendasse, A. 2009. Residual variance estimation in machine learning. Neurocomputing, 72(16–18), 3692–3703. https://doi.org/10.1016/j.neucom.2009.07.004
Mendez, G., & Lohr, S. 2011. Estimating residual variance in random forest regression. Computational Statistics & Data Analysis, 55(11), 2937–2950. https://doi.org/10.1016/j.csda.2011.04.022
Octavanny, M. A. D., Budiantara, I. N., Kuswanto, H., & Rahmawati, D. P. 2020. Nonparametric Regression Model for Longitudinal Data with Mixed Truncated Spline and Fourier Series. Abstract and Applied Analysis, 2020, 1–11. https://doi.org/10.1155/2020/4710745
Purnomo, J., Budiantara, I., & Fitriasari, K. 2008. Weight estimation using generalized moving average. IPTEK The Journal for Techhnology and Science, 19(4).
Rice, J. 1984. Bandwidth Choice for Nonparametric Regression. The Annals of Statistics, 12(4), 1215–1230. https://doi.org/10.1214/aos/1176346788
Rohatgi, V. . 1976. An Introduction to probability theory and mathematical statistics. USA: john Wiley & Sons Inc.
Serfling, R. J. 1980. Approximation theorems of mathematical statistics. USA: john Wiley & Sons Inc.
Spokoiny, V. 2002. Variance Estimation for High-Dimensional Regression Models. Journal of Multivariate Analysis, 82(1), 111–133. https://doi.org/10.1006/jmva.2001.2023
Tong, T., & Wang, Y. 2005. Estimating residual variance in nonparametric regression using least squares. Biometrika, 92(4), 821–830. https://doi.org/10.1093/biomet/92.4.821
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2021 Author and publisher
This work is licensed under a Creative Commons Attribution 4.0 International License.
This work is licensed under a Creative Commons Attribution 4.0 International License.
Jurnal Matematika, Statistika dan Komputasi is an Open Access journal, all articles are distributed under the terms of the Creative Commons Attribution License, allowing third parties to copy and redistribute the material in any medium or format, transform, and build upon the material, provided the original work is properly cited and states its license. This license allows authors and readers to use all articles, data sets, graphics and appendices in data mining applications, search engines, web sites, blogs and other platforms by providing appropriate reference.