HApsix computes psi function using Hampel proposal times x




psiHAx =HApsix(u, ctuning) Plot of psi(x) multiplied by x.


expand all

  • Plot of psi(x) multiplied by x.
  • Obtain bottom panel of Figure 11.10. p. 375 of Hoaglin et al. (1987)

    ylabel(' Hampel $\psi(x) \times x $','Interpreter','Latex')
    Click here for the graphical output of this example (link to Ro.S.A. website).

    Input Arguments

    expand all

    u — scaled residuals or Mahalanobis distances. Vector.

    n x 1 vector containing residuals or Mahalanobis distances for the n units of the sample

    Data Types: single| double

    ctuning — tuning parameters. Scalar or Vector.

    Scalar or vector of length 4 which specifies the value of the tuning constant c (scalar greater than 0 which controls the robustness/efficiency of the estimator) and the prefixed values of paramters a, b, c ctuning(1) = tuning constant which will multiply parameters a, b and c of Hampel rho (psi) function ctuning(2) = paramter a of Hampel rho (psi) function ctuning(3) = paramter b of Hampel rho (psi) function ctuning(4) = paramter c of Hampel rho (psi) function Remark: if length(ctuning)==1 values of a, b and c will be set to a=2*ctuning b=4*ctuning c=4*ctuning With these choices, if ctuning=1 the resulting influence function is nearly identical to the biweight with parameter 8.

    Data Types: single| double

    More About

    expand all

    Additional Details

    Function HApsix transforms vector u as follows \[ HApsi(u) = \left\{ \begin{array}{cc} u^2 & |u| <= a \\ a \times sign(u)u & a <= |u| < b \\ a \frac{c-|u|}{c-b} \times sign(u)\times u & b <= |u| < c \\ 0 & |u| >= c \end{array} \right. \]

    where $a$= ctuning(2) *ctuning(1).

    $b$= ctuning(3) *ctuning(1).

    $c$= ctuning(4) *ctuning(1).

    The default (if input ctuning is a scalar) is $a$= 2*ctuning.

    $b$= 4*ctuning.

    $c$= 8*ctuning.

    It is necessary to have 0 <= $a$ <= $b$ <= $c$


    Hoaglin, D.C., Mosteller, F., Tukey, J.W. (1982), "Understanding Robust and Exploratory Data Analysis", Wiley, New York.

    This page has been automatically generated by our routine publishFS