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
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
|
test_that("extractFDAFeatures", {
methods = list("UVVIS" = extractFDAMultiResFeatures(),
"NIR" = extractFDAFourier())
t = extractFDAFeatures(fuelsubset.task, feat.methods = methods)
# check output data
df = getTaskData(t$task)
expect_s3_class(df, "data.frame")
expect_equal(nrow(df), 129L)
expect_subset(colnames(df), c(paste0("NIR.phase.", seq_len(231)),
paste0("UVVIS.multires.", seq_len(9)), "heatan", "h20"))
})
test_that("extractFeatures multiple times", {
methods = list("UVVIS" = extractFDAMultiResFeatures(),
"UVVIS" = extractFDAFourier(),
"NIR" = extractFDAMultiResFeatures())
t = extractFDAFeatures(fuelsubset.task, feat.methods = methods)
# check output data
df = getTaskData(t$task)
expect_class(df, "data.frame")
expect_true(nrow(df) == 129L)
expect_true(ncol(df) == 154L)
expect_subset(colnames(df), c("heatan", "h20", paste0("UVVIS.phase.",
seq_len(134)),
paste0("NIR.multires.", seq_len(9)), paste0("UVVIS.multires.",
seq_len(9))))
methods = list("all" = extractFDAMultiResFeatures(), "all" = extractFDAFourier())
t = extractFDAFeatures(fuelsubset.task, feat.methods = methods)
# check output data
df = getTaskData(t$task)
expect_s3_class(df, "data.frame")
expect_true(nrow(df) == 129L)
expect_true(ncol(df) == 385L)
expect_subset(colnames(df),
c("heatan", "h20",
paste0("UVVIS.multires.", seq_len(9)), paste0("NIR.multires.", seq_len(9)),
paste0("UVVIS.phase.", seq_len(134)), paste0("NIR.phase.", seq_len(231))))
})
test_that("extractFDAFeatures colnames work", {
methods = list("NIR" = extractFDAFourier())
t = subsetTask(fuelsubset.task, subset = 1:30)
t2 = extractFDAFeatures(t, feat.methods = methods)
cn = getTaskFeatureNames(t2$task)
expect_match(setdiff(cn, "h2o"), regexp = "[NIR.phase]", all = TRUE)
})
test_that("Wrong methods yield errors", {
t = subsetTask(fuelsubset.task, subset = 1:2)
wrng1 = function() {
lrn = function(data, target, col, vals = NULL) {
1
}
makeExtractFDAFeatMethod(learn = lrn, reextract = lrn,
par.set = makeParamSet())
}
expect_error(extractFDAFeatures(t, feat.methods = list("NIR" = wrng1())),
"feat.method needs to return")
wrng2 = function() {
lrn = function(data) {
data[, 1]
}
makeExtractFDAFeatMethod(learn = lrn, reextract = lrn,
par.set = makeParamSet())
}
expect_error(extractFDAFeatures(t, feat.methods = list("NIR" = wrng2())),
"Must have formal arguments")
wrng3 = function() {
lrn = function(data, target, col, vals = NULL) {
data.frame(1)
}
makeExtractFDAFeatMethod(z = lrn, rz = lrn)
}
expect_error(extractFDAFeatures(t, feat.methods = list("NIR" = wrng3())),
"unused arguments")
})
test_that("extractFDAFeatures colnames work", {
methods = list("NIR" = extractFDAFourier())
t = subsetTask(fuelsubset.task, subset = 1)
t2 = extractFDAFeatures(t, feat.methods = methods)
cn = getTaskFeatureNames(t2$task)
expect_match(setdiff(cn, "h2o"), regexp = "[NIR.phase]", all = TRUE)
})
test_that("extractFDAFeaturesDesc", {
methods = list("UVVIS" = extractFDAMultiResFeatures(),
"NIR" = extractFDAFourier())
t = extractFDAFeatures(fuelsubset.task, feat.methods = methods)
# check desc
expect_s3_class(t$desc, "extractFDAFeatDesc")
expect_subset(t$desc$coln, c(getTaskFeatureNames(fuelsubset.task),
getTaskTargetNames(fuelsubset.task)))
expect_subset(t$desc$target, getTaskTargetNames(fuelsubset.task))
expect_subset(unique(t$desc$colclasses), choices = c("numeric", "matrix"))
expect_list(t$desc$extractFDAFeat)
expect_list(t$desc$extractFDAFeat$UVVIS$extractor.vals)
expect_function(t$desc$extractFDAFeat$UVVIS$reextract)
expect_list(t$desc$extractFDAFeat$NIR$extractor.vals)
expect_function(t$desc$extractFDAFeat$NIR$reextract)
})
test_that("extractFDAFeatures task equal data.frame", {
# check data.frame output equal to task's data output
gp.subset = subsetTask(gunpoint.task, features = 1L)
fm = list("fd" = extractFDAFourier(trafo.coeff = "amplitude"))
t2 = extractFDAFeatures(gp.subset, feat.methods = fm)
gp.desc = getTaskDesc(gp.subset)
t3 = extractFDAFeatures(getTaskData(gp.subset, functionals.as = "matrix"),
target = "X1", feat.methods = fm)
expect_identical(getTaskData(t2$task), t3$data)
expect_equal(t2$desc, t3$desc)
expect_equal(t2$desc$extractFDAFeat$fd$extractor.vals$trafo.coeff, "amplitude")
expect_error(extractFDAFeatures(gp.subset,
feat.methods = list("fd" = extractFDAFourier(),
"fd2" = extractFDAMultiResFeatures())), regexp = "Must be a subset of")
})
test_that("reextractFDAFeatures", {
gp.subset = subsetTask(gunpoint.task, features = 1L)
fm = list("fd" = extractFDAFourier(trafo.coeff = "amplitude"))
t3 = extractFDAFeatures(gp.subset, feat.methods = fm)
t4 = reextractFDAFeatures(gp.subset, t3$desc)
expect_equal(getTaskFeatureNames(t3$task), getTaskFeatureNames(t4))
expect_equal(t3$desc$target, getTaskTargetNames(t4))
expect_equal(dim(getTaskData(t3$task)), dim(getTaskData(t4)))
})
test_that("extract reextract feat.methods all", {
fm2 = list("all" = extractFDAFourier(trafo.coeff = "amplitude"))
t3 = extractFDAFeatures(fuelsubset.task, feat.methods = fm2)
t4 = reextractFDAFeatures(fuelsubset.task, t3$desc)
expect_equal(getTaskFeatureNames(t3$task), getTaskFeatureNames(t4))
expect_equal(t3$desc$target, getTaskTargetNames(t4))
expect_equal(dim(getTaskData(t3$task)), dim(getTaskData(t4)))
})
test_that("extract and reextract have correct args", {
lrn = makeExtractFDAFeatsWrapper("regr.rpart",
feat.methods = list("all" = extractFDAFourier()))
mod = train(setHyperPars(lrn, trafo.coeff = "amplitude"),
subsetTask(fuelsubset.task, subset = 1:20))
prd = predict(mod, subsetTask(fuelsubset.task, subset = 21:40))
expect_equal(mod$learner.model$control$extractFDAFeat$UVVIS$extractor.vals$trafo.coeff,
"amplitude")
expect_equal(mod$learner.model$control$extractFDAFeat$NIR$extractor.vals$trafo.coeff,
"amplitude")
})
|
