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
|
test_that("removeConstantFeatures", {
data = data.frame(
a = c(1L, 2L),
b = as.factor(1:2),
c = c("a", "b"),
d = as.factor(c(TRUE, FALSE)),
e = c(NA, 1),
f = c(1, NA),
g = c(1, 1),
n = c(0, 1 - 0.7 - 0.3),
m = as.double(c(NA, NA)), # only missings are supported?
target = as.factor(1:2),
stringsAsFactors = TRUE
)
data = data[c(rep(1, 9), 2), ]
data$safe = seq_row(data)
task = makeClassifTask("test", data = data, target = "target")
res1 = getTaskData(removeConstantFeatures(task, perc = 0.1, dont.rm = "g"))
res2 = removeConstantFeatures(getTaskData(task), perc = 0.1, dont.rm = c("g", "target"))
expect_equal(colnames(res1), c("g", "target", "safe"))
expect_equal(res1, res2)
res = getTaskData(removeConstantFeatures(task, na.ignore = TRUE))
expect_equal(colnames(res), c("a", "b", "c", "d", "target", "safe"))
res = getTaskData(removeConstantFeatures(task, wrap.tol = 0, na.ignore = TRUE))
expect_true(setequal(colnames(res), c("a", "b", "c", "d", "target", "safe", "n")))
res = getTaskData(removeConstantFeatures(task, na.ignore = FALSE))
expect_true(setequal(names(res), c("a", "b", "c", "d", "e", "f", "target", "safe")))
res = getTaskData(removeConstantFeatures(task, na.ignore = FALSE, perc = 0.2))
expect_equal(colnames(res), c("target", "safe"))
data = dropNamed(data, c("e", "f", "m"))
data$target = 1
data$noise = rnorm(nrow(data))
lrn = makeLearner("regr.lm")
lrn = makeRemoveConstantFeaturesWrapper(lrn, perc = 0.1)
task = makeRegrTask(data = data, target = "target")
model = train(lrn, task)
mod = getLearnerModel(model)$learner.model
expect_true(setequal(names(coef(mod)), c("(Intercept)", "noise", "safe")))
})
|
