| Library: | kalman |
| See also: | kemitor |
| Quantlet: | epscontnorm | |
| Description: | Produces T i.i.d. Variates from an eps-contamination Model P= (1-eps) N(mid,Cid) + eps K with K=N(mcont,Ccont) if DirNorm ==0 with K=dirac(mcont) if DirNorm == -1 with K=dirac( +/- mcont) if DirNorm == 1 |
| Usage: | {X,Ind}= epscontnorm(T,eps,mid,Cid,mcont,Ccont,DirNorm) | |
| Input: | ||
| T | number of Variates to be generated | |
| eps (contamination radius) | numeric in [0,1) | |
| mid (ideal mean) | n x 1 vector | |
| Cid (ideal Covariance) | n x n matrix | |
| mcont (contaminating mean) | n x 1 vector | |
| Ccont (contaminating Covariance) | n x n matrix | |
| DirNorm (decides which type of contaminating distribution K is taken) | K=N(mcont,Ccont) if DirNorm ==0 K=dirac(mcont) if DirNorm == -1 K=dirac( +/- mcont) if DirNorm == 1 | |
| Output: | ||
| X | T x n matrix of generated Variates, T is the number of generated Variates, n is the dimension of generated Variates | |
| Ind | T Indicator of contamination ( 0 = no cont. 1 = cont.) | |
library("kalman")
library("plot")
randomize(0)
T = 100
eps = 0.1
mid=#(0,0)
Cid = #(2,1)~#(1,1)
mcont=#(3,3)
Ccont = #(3,0)~#(0,0.2)
erg=epscontnorm(T,eps,mid,Cid,mcont,Ccont,0)
color=2*erg.Ind+2 ; red(4) if clipped green(2) else
data=erg.X
setmaskp(data,color, 3, 8) ; rest :default
disp = createdisplay(1,1)
show(disp,1,1,data)
Generates 100 observations of
(1-eps) N(0,Cid)+eps N(mcont,Ccont)
and plots the contaminated data in red, the rest in green