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
43
44
45
46
47
48
49
50
51
52
53
|
### Small Test Datasets (all kinds odd/even, constant/regular/outlier):
D <- within(list(), {
## n = 0,1,2,3 :
x0 <- numeric(0)
x1 <- 3
x1I <- Inf
x2 <- 1:2
x2I <- c(-Inf, 9)
xII <- c(-Inf, Inf)
x3 <- c(1:2,10)
x3I <- c(-Inf, 9,11)
x3.2I <- c(-Inf, 9, Inf)
## constant (0 mad) + 0--2 outliers
xC <- rep(1, 12)
xC. <- rep(1, 11)
xC1 <- c(xC, 10)
xC1. <- c(xC., 10)
xC2 <- c(xC1, 100)
xC2. <- c(xC1., 100)
## "uniform" + 0--2 outliers
y <- 1:10
y. <- 1:11
y1 <- c(y, 100)
y1. <- c(y., 100)
y2 <- c(y1, 1000)
y2. <- c(y1., 1000)
yI <- c(y1, Inf)
yI. <- c(y1., Inf)
})
smallD <- D[order(lengths(D))]
rm(D)
## Constructor of such "stylized" small data with large ('M') values / outliers:
mk3Mx <- function(M, ngood = 10, left = floor(ngood/3)) {
stopifnot(is.numeric(ngood), ngood >= 3,
is.numeric(M), length(M) == 1L, M >= 1000,
is.numeric(left), 0 <= left, left <= ngood)
right <- ngood-left
res <- list(
c(rep(-M, left), seq_len(ngood - 1L), rep(M, right)) # < 50% "good"
, c(rep(-M, left), seq_len(ngood ), rep(M, right)) # half "good"
, c(rep(-M, left), seq_len(ngood + 1L), rep(M, right)) # > 50% "good"
)
nM <- gsub("[-+]", "", formatC(M, digits=2, width=1))
names(res) <- paste0("M", nM,"_n", c("m1", "eq", "p1"))
res
}
## The one that works for a *vector* M:
mkMx <- function(M, ngood = 10, left = floor(ngood/3)) unlist(lapply(M, mk3Mx), recursive=FALSE)
|
