Improvement of Boundary Bias in Nonparametric Regression via Twicing Technique

  • Jo, Jae-Keun (Department of Computer Science and Statistics, Kyungsung University)
  • Published : 1997.08.01


In this paper, twicing technique for the improvement of asymptotic boundary bias in nonparametric regression is considered. Asymptotic mean squared errors of the nonparametric regression estimators are derived at the boundary region by twicing the Nadaraya-Waston and local linear smoothing. Asymptotic biases of the resulting estimators are of order$h^2$and$h^4$ respectively.



  1. Canadian Journal of Statistics v.23 Computationally Efficient Classes of Higher-order Kernel Functions Abdous, B.
  2. Local Polynomial Modelling and Its Applications Fan, J.;Gijbels, I.
  3. Applied Nonparametric Regression Hardle, W.
  4. Statistics and Computing v.3 Simple Boundary Correction for Kernel Density Estimation Jones, M.C.
  5. Journal of the Korean Statistical Society v.25 Ona Transformation Technique for Nonparametric Regression Kim, W.C.;Park, B.U.
  6. In Smoothing Techniques for Curve Estimation Some Comments on the Asymptotic Behaviour of Robust Smoothers Stuetzle, W.;Mittal, Y.;T. Gasser(ed.);M. Rosenblatt(ed.)
  7. Kernel Smoothing Wand, M.P.;Jones, M.C.