1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
|
library(robustbase)
## minimal testing only
data(ruspini, package = "cluster")
rub1 <- covOGK(ruspini, 1, scaleTau2, covGK, hard.rejection, consistency=FALSE)
rub2 <- covOGK(ruspini, 2, scaleTau2, covGK, hard.rejection, consistency=FALSE)
AE <- function(x,y) all.equal(x,y, tolerance = 2e-15)
## The following test is already fulfilled by Kjell Konis' original code:
stopifnot(AE(c(rub1$wcov)[c(1,3:4)],
c(917.99893333333, 94.9232, 2340.319288888888)),
all.equal(rub1$wcov, rub2$wcov, tolerance=0)
,
AE(c(rub1$cov)[c(1,3:4)],
c(923.5774514441657, 91.5385216376565, 2342.4556232436971))
,
AE(c(rub2$cov)[c(1,3:4)],
c(927.2465953711782, 91.8009184487779, 2346.5790105548940))
)
data(milk)
cM1 <- covOGK(milk, 1, sigmamu = scaleTau2, weight.fn = hard.rejection)
cM2 <- covOGK(milk, 2, sigmamu = scaleTau2, weight.fn = hard.rejection)
symnum(cov2cor(cM1 $cov))
symnum(cov2cor(cM2 $cov))
symnum(cov2cor(cM1 $wcov))
symnum(cov2cor(cM2 $wcov))
cMQn <- covOGK(milk, sigmamu = s_Qn, weight.fn = hard.rejection)
cMSn <- covOGK(milk, sigmamu = s_Sn, weight.fn = hard.rejection)
cMiqr <- covOGK(milk, sigmamu = s_IQR, weight.fn = hard.rejection)
cMmad <- covOGK(milk, sigmamu = s_mad, weight.fn = hard.rejection)
as.dist(round(cov2cor(cMQn$wcov), 3))
as.dist(round(cov2cor(cMSn$wcov), 3))
as.dist(round(cov2cor(cMiqr$wcov), 3))
as.dist(round(cov2cor(cMmad$wcov), 3))
cat('Time elapsed: ', proc.time(),'\n') # for ``statistical reasons''
|
