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
|
context("learners_all_regr")
test_that("learners work: regr ", {
# because of missing rJava for bartMachine
skip_on_os("windows")
requirePackagesOrSkip("crs", default.method = "load")
# settings to make learners faster and deal with small data size
hyperpars = list(
regr.km = list(nugget = 0.01),
regr.cforest = list(mtry = 1L, minsplit = 1, minbucket = 1),
regr.bartMachine = list(verbose = FALSE, run_in_sample = FALSE,
# see above
replace_missing_data_with_x_j_bar = TRUE,
num_iterations_after_burn_in = 10L),
regr.nodeHarvest = list(nodes = 100L, nodesize = 5L),
regr.h2o.deeplearning = list(hidden = 2L),
regr.ranger = list(keep.inbag = TRUE)
)
# Create smaller task: dont use feature 2, it is nearly always 0, don't use
# feature 4, it is a factor variable
task = subsetTask(regr.task, subset = c(1:70),
features = getTaskFeatureNames(regr.task)[c(1, 3)])
# normal regr
lrns = listLearnersCustom(task, create = TRUE)
lapply(lrns, testThatLearnerParamDefaultsAreInParamSet)
lapply(lrns, testBasicLearnerProperties, task = task, hyperpars = hyperpars)
# regr with factors
lrns = listLearnersCustom(task, properties = "factors", create = TRUE)
lapply(lrns, testThatLearnerHandlesFactors, task = task,
hyperpars = hyperpars)
# regr with ordered factors
lrns = listLearnersCustom(task, properties = "ordered", create = TRUE)
lapply(lrns, testThatLearnerHandlesOrderedFactors, task = task,
hyperpars = hyperpars)
# regr with se
lrns = listLearnersCustom(task, properties = "se", create = TRUE)
lapply(lrns, testBasicLearnerProperties, task = task, hyperpars = hyperpars,
pred.type = "se")
# regr with weights
lrns = listLearnersCustom(task, properties = "weights", create = TRUE)
lapply(lrns, testThatLearnerRespectsWeights, hyperpars = hyperpars,
task = task, train.inds = 1:70, test.inds = 1:70, weights = rep(c(1, 5),
length.out = 70),
pred.type = "response", get.pred.fun = getPredictionResponse)
# regr with missing
lrns = listLearnersCustom(task, properties = "missings", create = TRUE)
lapply(lrns, testThatLearnerHandlesMissings, task = task,
hyperpars = hyperpars)
# regr variable importance
lrns = listLearnersCustom(task, properties = "featimp", create = TRUE)
lapply(lrns, testThatLearnerCanCalculateImportance, task = task,
hyperpars = hyperpars)
# regr with oobpreds
lrns = listLearnersCustom(task, properties = "oobpreds", create = TRUE)
lapply(lrns, testThatGetOOBPredsWorks, task = task)
# regr with only one feature
min.task = makeRegrTask("oneCol", data.frame(x = 1:10, y = 1:10),
target = "y")
lrns = listLearnersCustom(min.task, create = TRUE)
# regr.gbm: Meaningfull error about too small dataset
# regr.cforest: Error in model@fit(data, ...) : fraction of 0.000000 is too small
# regr.nodeHarvest: Error in ZRULES[[1]] : subscript out of bounds
# others: see learners_all_classif and random errors
not.working = c(
"regr.cforest",
"regr.cvglmnet",
"regr.evtree",
"regr.frbs",
"regr.gbm",
"regr.glmnet",
"regr.laGP",
"regr.nodeHarvest",
"regr.slim")
lrns = lrns[extractSubList(lrns, "id", simplify = TRUE) %nin% not.working]
lapply(lrns, testBasicLearnerProperties, task = min.task,
hyperpars = hyperpars)
})
|
