A new Robust algorithm for penalized regression splines based on mode-estimation

Document Type : Research Paper


1 Department of Business Administration, College of Business, King Khaled University, KSA

2 Department of Statistics, Mathematics and Insurance, Faculty of Commerce, Alexandria University, Egypt


The main purpose of present article is proposed an effective method for robust fitting penalized regression splines models. According to such a context a comparative analysis with two common robust techniques, M-type estimator, S-type estimator, and non-robust least squares (LS) for penalized regression splines (PRS) has been implemented. Because the penalized regression splines are recently a common approach to smoothing noisy data for its simplicity, efficiency, and significantly reducing disturbance of outliers and its flexibility in monitoring nonlinear data trends. In many cases, it is difficult to determine the most suitable form and a way of designing a data is needed when faced with many smoothing problems. The executing aspects of fitting precision and robustness of the four estimators have a thorough evaluation of their performance on R codes. A comparative analysis demonstrates that the proposed method can resist the noise effect in both simulated and real data examples compared to other robust estimators with different combinations of contamination. These findings are used as guidance for finding a specific method to pulsing smoothing noisy data


[1] E.J. Wegman, and I.W. Wright, Splines in statistics, J. Amer. Statist. Assoc.78(382) (1983) 351-365.
[2] S.N. Wood, Modelling and smoothing parameter estimation with multiple quadratic penalties. J. Royal Statist. Soc.: Ser. B   62(2) (2000) 413-428.
[3] Y. J. Kim and C. Gu, Smoothing spline Gaussian regression: more scalable computation via efficient approximation, J. Royal Statist. Soc.: Ser. B   66(2) (2004) 337-356.
[4] F. O’Sullivan, A statistical perspective on ill-posed inverse problems, Statist. Sci. (1986) 502-518.
[5] C. Kelly and J. Rice, Monotone smoothing with application to dose-response curves and the assessment of synergism, Biometrics (1990) 1071-1085.
[6] B.D. Marx, and P.H. Eilers, Direct generalized additive modeling with penalized likelihood,  Comput. Statist. Data Anal. 28(2) (1998) 193-209.
[7] M.P. Wand, Miscellanea. On the optimal amount of smoothing in penalized spline regression. Biometrika 86(4) (1999) 936-940.
[8] D. Ruppert, M.P. Wand, and R.J. Carroll, Semiparametric Regression, Cambridge University Press, 2003.
[9] S. Wold, M. Sjstrm, and L. Eriksson, . PLS-regression: a basic tool of chemometrics, Chemomet. Intel. Lab. Syst. 58(2) (2001)109-130.
[10] K. Tharmaratnam, G. Claeskens, C. Croux, and M. Salibin-Barrera, S-estimation for penalized regression splines, J. Comput. Graphical Statist. 19(3) (2010) 609-625.
[11] D.D. Cox, Asymptotic for M-type smoothing splines, Ann. Statist. (1983) 530-551.
[12] T.C. Lee, and H.S. Oh, Robust penalized regression splines fitting with application to additive mixed modeling, Comput. Statist. 22(1) (2007) 159-171.
[13] P.H. Eilers, and B.D. Marx, Flexible smoothing with B-splines and penalties, Statist. Sci. (1996) 89-102.
[14] B. Wang, W. Shi, and Z. Miao, Comparative analysis for robust penalized spline smoothing methods, Math. Prob. Engin. 2014 (2014).
[15] P.J. Huber, Robust Statistics, New York, John Wiley and Sons, 1981.
[16] P. Rousseeuw and V. Yohai, Robust Regression by Means of S-type Estimators, Robust and nonlinear time series analysis (pp. 256-272), Springer, New York, NY, 1984.
[17] M. Salibian-Barrera, and V. Yohai, A fast algorithm for S-type regression estimates, J. Comput. Graphical Statist. 15(2) (2006) 414-427.
[18] H. S. Oh, D. Nychka, T. Brown, and P. Charbonneau, Period analysis of variable stars by robust smoothing. Journal of the Royal Statistical Society: Series C (Applied Statistics), 53(1),(2004) 15-30.
[19] R. Finger, Investigating the performance of different estimation techniques for crop yield data analysis in crop insurance applications. Agricultural Economics, 44(2),(2013) 217-230.
[20] C. L. Mallows. More comments on Cp. Technometrics, 37(4),(1995) 362-372.
[21] D. Ruppert, M. P. Wand, and R. J. Carroll, . Semiparametric regression during 20032007. Electronic journal of statistics, 3,(2009) 1193.
[22] N. R. Draper, and H. Smith. Applied regression analysis (Vol. 326). John Wiley and Sons(1998).
[23] T. Krivobokova, Theoretical and practical aspects of penalized spline smoothing, 2006).
[24] P. Craven and G. Wahba, Smoothing noisy data with spline functions, Numerische Math. 31(4) (1978) 377-403.
[25] C.M. Hurvich, J.S. Simonoff, and C.L. Tsai, Smoothing parameter selection in nonparametric regression using an improved Akaike information criterion. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 60(2 (1998) 271-293.
[26] R.R. Wilcox, Introduction to Robust Estimation and Hypothesis Testing, Academic Press, 2011.
[27] W. Zhao, R. Zhang, J. Liu, and Y. Lv, Robust and efficient variable selection for semiparametric partially linear varying coefficient model based on modal regression, Ann.  Inst. Statist. Math. 66(1) (2014) 165-191.
[28] W. Yao and L. Li, A new regression model: modal linear regression, Scand. J. Statist. 41(3) (2014) 656-671.
[29] B.W. Silverman, Density Estimation for Statistics and Data Analysis, Vol. 26, CRC Press, 1986)
[30] S. Wglarczyk, Kernel density estimation and its application, ITM Web Conf. (Vol. 23). EDP Sciences, 2018.
[31] E. Cantoni and E. Ronchetti, Resistant selection of the smoothing parameter for smoothing splines. Statist. Comput. 11(2) (2001) 141-146.
[32] W. Yao, B. G. Lindsay, and R. Li, Local modal regression, J. Nonparametric Statist. 24(3) (2012) 647-663.
Volume 12, Issue 1
May 2021
Pages 1037-1055
  • Receive Date: 31 October 2020
  • Revise Date: 27 December 2020
  • Accept Date: 13 January 2021