tests/testthat/test_base_coltypes.R
In mlr-org/automlr: Automatic Machine Learning in R
# Test that feature types influence the presence and absence of learners, wrapper parameters and pseudoparameters
# in the right way.
context("coltypes")
# the subset of learners that can use the given task is chosen
test_that("the correct learner is automatically chosen", {
nlr = NumericsLearner
nlr$searchspace[[1]] = sp("int1", "int", c(0, 10), req = quote(2 %in% c(1, 2, 3)))
# test that '2 %in% c(1, 2, 3)' is removed since it is always true.
l = blt(list(nlr), NumericsTask)
expect_null(getpars(l)$NumericsLearner.int1$requires)
# only numericslearner remains for numericstask
expect_message(l <- blt(list(NumericsLearner, FactorsLearner, OrderedsLearner), NumericsTask),
"Learner can not handle feature types", all = TRUE)
checkLearnerBehaviour(l, NumericsTask, list(NumericsLearner.int1 = 1), "NumericsLearner", list(int1 = 1), list())
# both numericslearner and missingsnumericslearner work on numericstask
l = blt(list(NumericsLearner, MissingsNumericsLearner), NumericsTask)
checkLearnerBehaviour(l, NumericsTask, list(selected.learner = "MissingsNumericsLearner"),
"MissingsNumericsLearner", list(), list())
# only missingsnumericslearner works on missingsnumericstask, int1 must be set since missings are present
expect_message(l <- blt(list(NumericsLearner, MissingsNumericsLearner), MissingsNumericsTask),
"Learner can not handle the task", all = TRUE)
checkLearnerBehaviour(l, MissingsNumericsTask, list(MissingsNumericsLearner.int1 = 2),
"MissingsNumericsLearner", list(int1 = 2), list())
})
# test that automlr.has.missings works as it should, is present only when needed
test_that("parameters are fed to the correct learner / wrapper", {
l = buildLearners(list(NumericsLearner, MissingsNumericsLearner, pWW(nimp1), pWW(nimp2)), NumericsTask, verbosity = 5)
# numericslearner for numericstask: wrappers don't do anything.
checkLearnerBehaviour(l, NumericsTask,
list(selected.learner = "NumericsLearner", NumericsLearner.int1 = 3),
"NumericsLearner", list(int1 = 3), list())
# numericslearner for missingsnumericstask: NARemover automatically removes NAs
# also watch changed order of wrappers
l = buildLearners(list(NumericsLearner, MissingsNumericsLearner, pWW(nimp1), pWW(nimp2)), MissingsNumericsTask, verbosity = 6)
checkLearnerBehaviour(l, MissingsNumericsTask,
list(selected.learner = "NumericsLearner", NumericsLearner.int1 = 3, automlr.wimputing.numerics = "numimputer1"),
"NumericsLearner", list(int1 = 3),
numimputer1 = list())
checkLearnerBehaviour(l, MissingsNumericsTask,
list(selected.learner = "NumericsLearner", NumericsLearner.int1 = 3, automlr.wimputing.numerics = "numimputer2", multiplier = 2),
"NumericsLearner", list(int1 = 3), numimputer2 = list(multiplier = 2))
# using MissingsNumericsLearner has the same effect if automlr.remove.missings == TRUE
l = buildLearners(list(NumericsLearner, MissingsNumericsLearner, pWW(nimp1), pWW(nimp2), pWW(fimp1), pWW(fimp2), pWW(oimp1), pWW(oimp2)),
MissingsNumericsTask, verbosity = 6)
checkLearnerBehaviour(l, NumericsTask,
list(selected.learner = "MissingsNumericsLearner", automlr.impute = TRUE, automlr.wimputing.numerics = "numimputer2", multiplier = 2),
"MissingsNumericsLearner", list(), list(),
numimputer2 = list(multiplier = 2))
# with automlr.remove.missings == FALSE, MissingsNumericsLearner.int1 has true requirement.
l = buildLearners(list(NumericsLearner, MissingsNumericsLearner, pWW(nimp1), pWW(nimp2), pWW(fimp1), pWW(fimp2), pWW(oimp1), pWW(oimp2)),
MissingsNumericsTask, verbosity = 6)
checkLearnerBehaviour(l, NumericsTask,
list(selected.learner = "MissingsNumericsLearner", automlr.impute = FALSE, MissingsNumericsLearner.int1 = 9),
"MissingsNumericsLearner", list(int1 = 9), list())
})
# automlr.impute and automlr.wimputing are present in the right cases, and cause the right behaviour.
test_that("requirements using pseudoparameters behave as expected", {
# all learners and most removers
expect_warning(l <- buildLearners(list(
NumericsLearner, FactorsLearner, OrderedsLearner,
MissingsNumericsLearner, MissingsFactorsLearner, MissingsFactorsNumericsLearner,
FactorsNumericsLearner, AllLearner,
pWW(nimp1), pWW(nimp2), pWW(fimp1), pWW(fimp2), pWW(oimp1), pWW(oimp2),
pWW(fnconv1), pWW(onconv1), pWW(onconv2),
pWW(nfconv1), pWW(nfconv2), pWW(ofconv1), pWW(ofconv2),
pWW(noconv1), pWW(noconv2), pWW(foconv1), pWW(foconv2)), MissingsNumericsFactorsTask, verbosity = 6),
"different \\(but feasible\\) type 'cat' listed|has parameters .* not mentioned in search space", all = TRUE)
# the wrappers that can remove factors, or missings, are the expected ones
expect_set_equal(unlist(getpars(l)$automlr.wconverting.factors.to.ordered$values), c("foconv1", "foconv2"))
expect_set_equal(unlist(getpars(l)$automlr.wimputing.ordered$values), c("ordimputer1", "ordimputer2"))
# the numericslearner requires removal of factors and missings; parameters correctly determine which conversions are necessary.
pvs = list(selected.learner = "NumericsLearner", automlr.missing.indicators = TRUE, automlr.convert.before.impute = FALSE,
automlr.wimputing.factors = "factimputer1",
automlr.wimputing.numerics = "numimputer1",
reference.cat = FALSE, fimp.const = "NAx", NumericsLearner.int1 = 8)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "NumericsLearner", list(int1 = 8), list(),
factimputer1 = list(fimp.const = "NAx"),
numimputer1 = list(),
fnconv1 = list(reference.cat = FALSE))
pvx = namedList(getParamIds(getParamSet(l)), NA)
isFeasible(getParamSet(l), insert(pvx, pvs))
pvs = list(selected.learner = "NumericsLearner", automlr.missing.indicators = TRUE, automlr.convert.before.impute = TRUE,
automlr.wimputing.numerics = "numimputer2", multiplier = 1,
reference.cat = FALSE, NumericsLearner.int1 = 9)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "NumericsLearner", list(int1 = 9), list(),
fnconv1 = list(reference.cat = FALSE),
numimputer2 = list(multiplier = 1))
pvs = list(selected.learner = "NumericsLearner", automlr.missing.indicators = FALSE,
automlr.convert.before.impute = TRUE,
automlr.wimputing.numerics = "numimputer2", multiplier = 1,
reference.cat = FALSE, NumericsLearner.int1 = 9)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "NumericsLearner", list(int1 = 9), list(),
fnconv1 = list(reference.cat = FALSE),
numimputer2 = list(multiplier = 1))
# The learner that can deal with missings and factors adds parameter automlr.impute. The wimputing / wconverting parameter is only present if the corresponding conversion happens.
pvs = list(selected.learner = "MissingsFactorsNumericsLearner",
automlr.convert.factors = TRUE, automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.numerics = FALSE,
automlr.missing.indicators = TRUE, automlr.convert.before.impute = FALSE,
automlr.wimputing.factors = "factimputer1",
automlr.wimputing.numerics = "numimputer1",
reference.cat = FALSE, fimp.const = "NAx")
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "MissingsFactorsNumericsLearner", list(), list(),
factimputer1= list(fimp.const = "NAx"),
numimputer1 = list(),
fnconv1 = list(reference.cat = FALSE))
pvs = list(selected.learner = "MissingsFactorsNumericsLearner",
automlr.convert.factors = TRUE, automlr.impute = FALSE, automlr.convert = TRUE,
automlr.convert.numerics = TRUE,
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.factors = "nfconv2",
reference.cat = FALSE, conv2.numsplits = 2,
MissingsFactorsNumericsLearner.int1 = 9)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "MissingsFactorsNumericsLearner", list(int1 = 9, real1 = 10), list(),
fnconv1 = list(reference.cat = FALSE),
nfconv2 = list(conv2.numsplits = 2))
pvs = list(selected.learner = "MissingsFactorsNumericsLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.before.impute = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.factors = FALSE,
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.factors = "nfconv2",
conv2.numsplits = 2,
automlr.wimputing.factors = "factimputer1", fimp.const = "NAx",
MissingsFactorsNumericsLearner.bool1 = TRUE)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "MissingsFactorsNumericsLearner", list(bool1 = TRUE, real1 = 10), list(),
nfconv2 = list(conv2.numsplits = 2),
factimputer1 = list(fimp.const = "NAx"))
# AllLearner behaves essentially the same as MissingsFactorsNumericsLearner for the given task.
pvs = list(selected.learner = "AllLearner",
automlr.convert.factors = TRUE, automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.factors.to = "numerics",
automlr.convert.numerics = FALSE,
automlr.missing.indicators = TRUE, automlr.convert.before.impute = FALSE,
automlr.wimputing.factors = "factimputer1",
automlr.wimputing.numerics = "numimputer1",
reference.cat = FALSE, fimp.const = "NAx",
AllLearner.int1 = 9, AllLearner.real1 = 8)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "AllLearner", list(bool1 = TRUE, int1 = 9, real1 = 8), list(),
factimputer1= list(fimp.const = "NAx"),
numimputer1 = list(),
fnconv1 = list(reference.cat = FALSE))
pvs = list(selected.learner = "AllLearner",
automlr.convert.factors = TRUE, automlr.impute = FALSE, automlr.convert = TRUE,
automlr.convert.factors.to = "numerics",
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "factors",
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.factors = "nfconv2",
reference.cat = FALSE, conv2.numsplits = 2,
AllLearner.real1 = 10)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "AllLearner", list(int1 = 2, real1 = 10), list(),
fnconv1 = list(reference.cat = FALSE),
nfconv2 = list(conv2.numsplits = 2))
pvs = list(selected.learner = "AllLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.before.impute = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "factors",
automlr.convert.factors = FALSE,
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.factors = "nfconv2",
conv2.numsplits = 2,
automlr.wimputing.factors = "factimputer1", fimp.const = "NAx",
AllLearner.real1 = 10, AllLearner.int1 = 6)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "AllLearner", list(int1 = 6, real1 = 10), list(),
nfconv2 = list(conv2.numsplits = 2),
factimputer1 = list(fimp.const = "NAx"))
pvs = list(selected.learner = "AllLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.before.impute = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "ordered",
automlr.convert.factors = TRUE,
automlr.convert.factors.to = "ordered",
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.ordered = "noconv2",
automlr.wconverting.factors.to.ordered = "foconv1",
automlr.wimputing.ordered = "ordimputer1", oimp.const = "NAx",
AllLearner.real1 = 10, AllLearner.int1 = 6)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "AllLearner", list(int1 = 6, real1 = 10), list(),
foconv1 = list(), noconv2 = list(),
ordimputer1 = list(oimp.const = "NAx"))
pvs = list(selected.learner = "AllLearner",
automlr.impute = FALSE, automlr.convert = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "ordered",
automlr.convert.factors = TRUE,
automlr.convert.factors.to = "ordered",
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.ordered = "noconv2",
automlr.wconverting.factors.to.ordered = "foconv1",
AllLearner.real1 = 10, AllLearner.int1.AMLRFIX2 = 12)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "AllLearner", list(int1 = 12, real1 = 10), list(),
foconv1 = list(), noconv2 = list())
})
# automlr.impute and automlr.wimputing are present in the right cases, and cause the right behaviour.
test_that("factors are present whenever missing indicators are introduced", {
# all learners and most removers
expect_warning(l <- buildLearners(list(
NumericsLearner, FactorsLearner, OrderedsLearner,
MissingsNumericsLearner, MissingsFactorsLearner, MissingsFactorsNumericsLearner,
FactorsNumericsLearner, AllLearner,
pWW(nimp1), pWW(nimp2), pWW(fimp1), pWW(fimp2), pWW(oimp1), pWW(oimp2),
pWW(fnconv1), pWW(onconv1), pWW(onconv2),
pWW(nfconv1), pWW(nfconv2), pWW(ofconv1), pWW(ofconv2),
pWW(noconv1), pWW(noconv2), pWW(foconv1), pWW(foconv2)), MissingsNumericsTask, verbosity = 6),
"different \\(but feasible\\) type 'cat' listed|has parameters .* not mentioned in search space", all = TRUE)
pvs = list(selected.learner = "NumericsLearner", automlr.missing.indicators = TRUE, automlr.convert.before.impute = TRUE,
automlr.wimputing.numerics = "numimputer2", multiplier = 1,
reference.cat = FALSE, NumericsLearner.int1 = 9)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "NumericsLearner", list(int1 = 9), list(),
fnconv1 = list(reference.cat = FALSE),
numimputer2 = list(multiplier = 1))
pvs = list(selected.learner = "NumericsLearner", automlr.missing.indicators = TRUE,
automlr.convert.before.impute = TRUE,
automlr.wimputing.numerics = "numimputer2", multiplier = 1,
reference.cat = FALSE, NumericsLearner.int1 = 9)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "NumericsLearner", list(int1 = 9), list(),
fnconv1 = list(reference.cat = FALSE),
numimputer2 = list(multiplier = 1))
pvs = list(selected.learner = "NumericsLearner", automlr.missing.indicators = FALSE,
automlr.wimputing.numerics = "numimputer2", multiplier = 1,
NumericsLearner.int1 = 9)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "NumericsLearner", list(int1 = 9), list(),
numimputer2 = list(multiplier = 1))
# The learner that can deal with missings and factors adds parameter automlr.impute. The wimputing / wconverting parameter is only present if the corresponding conversion happens.
pvs = list(selected.learner = "MissingsFactorsNumericsLearner",
automlr.convert.factors = TRUE, automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.numerics = FALSE,
automlr.missing.indicators = TRUE, automlr.convert.before.impute = FALSE,
automlr.wimputing.numerics = "numimputer1",
reference.cat = FALSE)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "MissingsFactorsNumericsLearner", list(), list(),
numimputer1 = list(),
fnconv1 = list(reference.cat = FALSE))
pvs = list(selected.learner = "MissingsFactorsNumericsLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.numerics = FALSE,
automlr.missing.indicators = FALSE,
automlr.wimputing.numerics = "numimputer1")
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "MissingsFactorsNumericsLearner", list(), list(),
numimputer1 = list())
pvs = list(selected.learner = "MissingsFactorsNumericsLearner",
automlr.convert.factors = TRUE, automlr.impute = FALSE, automlr.convert = TRUE,
automlr.convert.numerics = TRUE,
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.factors = "nfconv2",
reference.cat = FALSE, conv2.numsplits = 2,
MissingsFactorsNumericsLearner.int1 = 9)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "MissingsFactorsNumericsLearner", list(int1 = 9, real1 = 10), list(),
fnconv1 = list(reference.cat = FALSE),
nfconv2 = list(conv2.numsplits = 2))
pvs = list(selected.learner = "MissingsFactorsNumericsLearner",
automlr.impute = FALSE, automlr.convert = TRUE,
automlr.convert.numerics = TRUE,
automlr.missing.indicators = FALSE,
automlr.wconverting.numerics.to.factors = "nfconv2",
conv2.numsplits = 2,
MissingsFactorsNumericsLearner.int1 = 9)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "MissingsFactorsNumericsLearner", list(int1 = 9, real1 = 10), list(),
nfconv2 = list(conv2.numsplits = 2))
pvs = list(selected.learner = "MissingsFactorsNumericsLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.before.impute = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.factors = FALSE,
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.factors = "nfconv2",
conv2.numsplits = 2,
automlr.wimputing.factors = "factimputer1", fimp.const = "NAx",
MissingsFactorsNumericsLearner.bool1 = TRUE)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "MissingsFactorsNumericsLearner", list(bool1 = TRUE, real1 = 10), list(),
nfconv2 = list(conv2.numsplits = 2),
factimputer1 = list(fimp.const = "NAx"))
pvs = list(selected.learner = "MissingsFactorsNumericsLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.before.impute = TRUE,
automlr.convert.numerics = TRUE,
automlr.missing.indicators = FALSE,
automlr.wconverting.numerics.to.factors = "nfconv2",
conv2.numsplits = 2,
automlr.wimputing.factors = "factimputer1", fimp.const = "NAx",
MissingsFactorsNumericsLearner.bool1 = TRUE)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "MissingsFactorsNumericsLearner", list(bool1 = TRUE, real1 = 10), list(),
nfconv2 = list(conv2.numsplits = 2),
factimputer1 = list(fimp.const = "NAx"))
# AllLearner behaves essentially the same as MissingsFactorsNumericsLearner for the given task.
pvs = list(selected.learner = "AllLearner",
automlr.convert.factors = TRUE, automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.factors.to = "numerics",
automlr.convert.numerics = FALSE,
automlr.missing.indicators = TRUE, automlr.convert.before.impute = FALSE,
automlr.wimputing.numerics = "numimputer1",
reference.cat = FALSE,
AllLearner.int1 = 9, AllLearner.real1 = 8)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(bool1 = TRUE, int1 = 9, real1 = 8), list(),
numimputer1 = list(),
fnconv1 = list(reference.cat = FALSE))
pvs = list(selected.learner = "AllLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.numerics = FALSE,
automlr.missing.indicators = FALSE,
automlr.wimputing.numerics = "numimputer1",
AllLearner.int1 = 9, AllLearner.real1 = 8)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(bool1 = TRUE, int1 = 9, real1 = 8), list(),
numimputer1 = list())
pvs = list(selected.learner = "AllLearner",
automlr.convert.factors = TRUE, automlr.impute = FALSE, automlr.convert = TRUE,
automlr.convert.factors.to = "numerics",
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "factors",
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.factors = "nfconv2",
reference.cat = FALSE, conv2.numsplits = 2,
AllLearner.real1 = 10)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(int1 = 2, real1 = 10), list(),
fnconv1 = list(reference.cat = FALSE),
nfconv2 = list(conv2.numsplits = 2))
pvs = list(selected.learner = "AllLearner",
automlr.impute = FALSE, automlr.convert = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "factors",
automlr.missing.indicators = FALSE,
automlr.wconverting.numerics.to.factors = "nfconv2",
conv2.numsplits = 2,
AllLearner.real1 = 10)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(int1 = 2, real1 = 10), list(),
nfconv2 = list(conv2.numsplits = 2))
pvs = list(selected.learner = "AllLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.before.impute = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "factors",
automlr.convert.factors = FALSE,
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.factors = "nfconv2",
conv2.numsplits = 2,
automlr.wimputing.factors = "factimputer1", fimp.const = "NAx",
AllLearner.real1 = 10, AllLearner.int1 = 6)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(int1 = 6, real1 = 10), list(),
nfconv2 = list(conv2.numsplits = 2),
factimputer1 = list(fimp.const = "NAx"))
pvs = list(selected.learner = "AllLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.before.impute = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "factors",
automlr.missing.indicators = FALSE,
automlr.wconverting.numerics.to.factors = "nfconv2",
conv2.numsplits = 2,
automlr.wimputing.factors = "factimputer1", fimp.const = "NAx",
AllLearner.real1 = 10, AllLearner.int1 = 6)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(int1 = 6, real1 = 10), list(),
nfconv2 = list(conv2.numsplits = 2),
factimputer1 = list(fimp.const = "NAx"))
pvs = list(selected.learner = "AllLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.before.impute = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "ordered",
automlr.convert.factors = TRUE,
automlr.convert.factors.to = "ordered",
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.ordered = "noconv2",
automlr.wconverting.factors.to.ordered = "foconv1",
automlr.wimputing.ordered = "ordimputer1", oimp.const = "NAx",
AllLearner.real1 = 10, AllLearner.int1 = 6)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(int1 = 6, real1 = 10), list(),
foconv1 = list(), noconv2 = list(),
ordimputer1 = list(oimp.const = "NAx"))
pvs = list(selected.learner = "AllLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.before.impute = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "ordered",
automlr.missing.indicators = FALSE,
automlr.wconverting.numerics.to.ordered = "noconv2",
automlr.wimputing.ordered = "ordimputer1", oimp.const = "NAx",
AllLearner.real1 = 10, AllLearner.int1 = 6)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(int1 = 6, real1 = 10), list(),
noconv2 = list(),
ordimputer1 = list(oimp.const = "NAx"))
pvs = list(selected.learner = "AllLearner",
automlr.impute = FALSE, automlr.convert = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "ordered",
automlr.convert.factors = TRUE,
automlr.convert.factors.to = "ordered",
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.ordered = "noconv2",
automlr.wconverting.factors.to.ordered = "foconv1",
AllLearner.real1 = 10, AllLearner.int1.AMLRFIX2 = 12)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(int1 = 12, real1 = 10), list(),
foconv1 = list(), noconv2 = list())
pvs = list(selected.learner = "AllLearner",
automlr.impute = FALSE, automlr.convert = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "ordered",
automlr.missing.indicators = FALSE,
automlr.wconverting.numerics.to.ordered = "noconv2",
AllLearner.real1 = 10, AllLearner.int1.AMLRFIX2 = 12)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(int1 = 12, real1 = 10), list(),
noconv2 = list())
})
test_that("targettype pseudoparameter is set as expected", {
nlr = NumericsLearner
nlr$searchspace = list(
sp("int1", "int", c(0, 0),
req = quote(automlr.targettype == "multiclass")),
sp("int1.AMLRFIX1", "int", c(1, 1),
req = quote(automlr.targettype == "twoclass")))
l = blt(list(nlr), NumericsTask)
checkLearnerBehaviour(l, NumericsTask, list(), "NumericsLearner",
list(int1 = 1), list())
tdat = getTaskData(NumericsTask)
tdat$fac.1 = factor(c("a", "b", "c", "b"))
NTask2 = makeClassifTask("NumericsTask2", tdat, "fac.1")
l = blt(list(nlr), NTask2)
checkLearnerBehaviour(l, NTask2, list(), "NumericsLearner",
list(int1 = 0), list())
})
mlr-org/automlr documentation built on May 23, 2019, 3:02 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
# Test that feature types influence the presence and absence of learners, wrapper parameters and pseudoparameters
# in the right way.
context("coltypes")
# the subset of learners that can use the given task is chosen
test_that("the correct learner is automatically chosen", {
nlr = NumericsLearner
nlr$searchspace[[1]] = sp("int1", "int", c(0, 10), req = quote(2 %in% c(1, 2, 3)))
# test that '2 %in% c(1, 2, 3)' is removed since it is always true.
l = blt(list(nlr), NumericsTask)
expect_null(getpars(l)$NumericsLearner.int1$requires)
# only numericslearner remains for numericstask
expect_message(l <- blt(list(NumericsLearner, FactorsLearner, OrderedsLearner), NumericsTask),
"Learner can not handle feature types", all = TRUE)
checkLearnerBehaviour(l, NumericsTask, list(NumericsLearner.int1 = 1), "NumericsLearner", list(int1 = 1), list())
# both numericslearner and missingsnumericslearner work on numericstask
l = blt(list(NumericsLearner, MissingsNumericsLearner), NumericsTask)
checkLearnerBehaviour(l, NumericsTask, list(selected.learner = "MissingsNumericsLearner"),
"MissingsNumericsLearner", list(), list())
# only missingsnumericslearner works on missingsnumericstask, int1 must be set since missings are present
expect_message(l <- blt(list(NumericsLearner, MissingsNumericsLearner), MissingsNumericsTask),
"Learner can not handle the task", all = TRUE)
checkLearnerBehaviour(l, MissingsNumericsTask, list(MissingsNumericsLearner.int1 = 2),
"MissingsNumericsLearner", list(int1 = 2), list())
})
# test that automlr.has.missings works as it should, is present only when needed
test_that("parameters are fed to the correct learner / wrapper", {
l = buildLearners(list(NumericsLearner, MissingsNumericsLearner, pWW(nimp1), pWW(nimp2)), NumericsTask, verbosity = 5)
# numericslearner for numericstask: wrappers don't do anything.
checkLearnerBehaviour(l, NumericsTask,
list(selected.learner = "NumericsLearner", NumericsLearner.int1 = 3),
"NumericsLearner", list(int1 = 3), list())
# numericslearner for missingsnumericstask: NARemover automatically removes NAs
# also watch changed order of wrappers
l = buildLearners(list(NumericsLearner, MissingsNumericsLearner, pWW(nimp1), pWW(nimp2)), MissingsNumericsTask, verbosity = 6)
checkLearnerBehaviour(l, MissingsNumericsTask,
list(selected.learner = "NumericsLearner", NumericsLearner.int1 = 3, automlr.wimputing.numerics = "numimputer1"),
"NumericsLearner", list(int1 = 3),
numimputer1 = list())
checkLearnerBehaviour(l, MissingsNumericsTask,
list(selected.learner = "NumericsLearner", NumericsLearner.int1 = 3, automlr.wimputing.numerics = "numimputer2", multiplier = 2),
"NumericsLearner", list(int1 = 3), numimputer2 = list(multiplier = 2))
# using MissingsNumericsLearner has the same effect if automlr.remove.missings == TRUE
l = buildLearners(list(NumericsLearner, MissingsNumericsLearner, pWW(nimp1), pWW(nimp2), pWW(fimp1), pWW(fimp2), pWW(oimp1), pWW(oimp2)),
MissingsNumericsTask, verbosity = 6)
checkLearnerBehaviour(l, NumericsTask,
list(selected.learner = "MissingsNumericsLearner", automlr.impute = TRUE, automlr.wimputing.numerics = "numimputer2", multiplier = 2),
"MissingsNumericsLearner", list(), list(),
numimputer2 = list(multiplier = 2))
# with automlr.remove.missings == FALSE, MissingsNumericsLearner.int1 has true requirement.
l = buildLearners(list(NumericsLearner, MissingsNumericsLearner, pWW(nimp1), pWW(nimp2), pWW(fimp1), pWW(fimp2), pWW(oimp1), pWW(oimp2)),
MissingsNumericsTask, verbosity = 6)
checkLearnerBehaviour(l, NumericsTask,
list(selected.learner = "MissingsNumericsLearner", automlr.impute = FALSE, MissingsNumericsLearner.int1 = 9),
"MissingsNumericsLearner", list(int1 = 9), list())
})
# automlr.impute and automlr.wimputing are present in the right cases, and cause the right behaviour.
test_that("requirements using pseudoparameters behave as expected", {
# all learners and most removers
expect_warning(l <- buildLearners(list(
NumericsLearner, FactorsLearner, OrderedsLearner,
MissingsNumericsLearner, MissingsFactorsLearner, MissingsFactorsNumericsLearner,
FactorsNumericsLearner, AllLearner,
pWW(nimp1), pWW(nimp2), pWW(fimp1), pWW(fimp2), pWW(oimp1), pWW(oimp2),
pWW(fnconv1), pWW(onconv1), pWW(onconv2),
pWW(nfconv1), pWW(nfconv2), pWW(ofconv1), pWW(ofconv2),
pWW(noconv1), pWW(noconv2), pWW(foconv1), pWW(foconv2)), MissingsNumericsFactorsTask, verbosity = 6),
"different \\(but feasible\\) type 'cat' listed|has parameters .* not mentioned in search space", all = TRUE)
# the wrappers that can remove factors, or missings, are the expected ones
expect_set_equal(unlist(getpars(l)$automlr.wconverting.factors.to.ordered$values), c("foconv1", "foconv2"))
expect_set_equal(unlist(getpars(l)$automlr.wimputing.ordered$values), c("ordimputer1", "ordimputer2"))
# the numericslearner requires removal of factors and missings; parameters correctly determine which conversions are necessary.
pvs = list(selected.learner = "NumericsLearner", automlr.missing.indicators = TRUE, automlr.convert.before.impute = FALSE,
automlr.wimputing.factors = "factimputer1",
automlr.wimputing.numerics = "numimputer1",
reference.cat = FALSE, fimp.const = "NAx", NumericsLearner.int1 = 8)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "NumericsLearner", list(int1 = 8), list(),
factimputer1 = list(fimp.const = "NAx"),
numimputer1 = list(),
fnconv1 = list(reference.cat = FALSE))
pvx = namedList(getParamIds(getParamSet(l)), NA)
isFeasible(getParamSet(l), insert(pvx, pvs))
pvs = list(selected.learner = "NumericsLearner", automlr.missing.indicators = TRUE, automlr.convert.before.impute = TRUE,
automlr.wimputing.numerics = "numimputer2", multiplier = 1,
reference.cat = FALSE, NumericsLearner.int1 = 9)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "NumericsLearner", list(int1 = 9), list(),
fnconv1 = list(reference.cat = FALSE),
numimputer2 = list(multiplier = 1))
pvs = list(selected.learner = "NumericsLearner", automlr.missing.indicators = FALSE,
automlr.convert.before.impute = TRUE,
automlr.wimputing.numerics = "numimputer2", multiplier = 1,
reference.cat = FALSE, NumericsLearner.int1 = 9)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "NumericsLearner", list(int1 = 9), list(),
fnconv1 = list(reference.cat = FALSE),
numimputer2 = list(multiplier = 1))
# The learner that can deal with missings and factors adds parameter automlr.impute. The wimputing / wconverting parameter is only present if the corresponding conversion happens.
pvs = list(selected.learner = "MissingsFactorsNumericsLearner",
automlr.convert.factors = TRUE, automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.numerics = FALSE,
automlr.missing.indicators = TRUE, automlr.convert.before.impute = FALSE,
automlr.wimputing.factors = "factimputer1",
automlr.wimputing.numerics = "numimputer1",
reference.cat = FALSE, fimp.const = "NAx")
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "MissingsFactorsNumericsLearner", list(), list(),
factimputer1= list(fimp.const = "NAx"),
numimputer1 = list(),
fnconv1 = list(reference.cat = FALSE))
pvs = list(selected.learner = "MissingsFactorsNumericsLearner",
automlr.convert.factors = TRUE, automlr.impute = FALSE, automlr.convert = TRUE,
automlr.convert.numerics = TRUE,
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.factors = "nfconv2",
reference.cat = FALSE, conv2.numsplits = 2,
MissingsFactorsNumericsLearner.int1 = 9)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "MissingsFactorsNumericsLearner", list(int1 = 9, real1 = 10), list(),
fnconv1 = list(reference.cat = FALSE),
nfconv2 = list(conv2.numsplits = 2))
pvs = list(selected.learner = "MissingsFactorsNumericsLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.before.impute = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.factors = FALSE,
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.factors = "nfconv2",
conv2.numsplits = 2,
automlr.wimputing.factors = "factimputer1", fimp.const = "NAx",
MissingsFactorsNumericsLearner.bool1 = TRUE)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "MissingsFactorsNumericsLearner", list(bool1 = TRUE, real1 = 10), list(),
nfconv2 = list(conv2.numsplits = 2),
factimputer1 = list(fimp.const = "NAx"))
# AllLearner behaves essentially the same as MissingsFactorsNumericsLearner for the given task.
pvs = list(selected.learner = "AllLearner",
automlr.convert.factors = TRUE, automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.factors.to = "numerics",
automlr.convert.numerics = FALSE,
automlr.missing.indicators = TRUE, automlr.convert.before.impute = FALSE,
automlr.wimputing.factors = "factimputer1",
automlr.wimputing.numerics = "numimputer1",
reference.cat = FALSE, fimp.const = "NAx",
AllLearner.int1 = 9, AllLearner.real1 = 8)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "AllLearner", list(bool1 = TRUE, int1 = 9, real1 = 8), list(),
factimputer1= list(fimp.const = "NAx"),
numimputer1 = list(),
fnconv1 = list(reference.cat = FALSE))
pvs = list(selected.learner = "AllLearner",
automlr.convert.factors = TRUE, automlr.impute = FALSE, automlr.convert = TRUE,
automlr.convert.factors.to = "numerics",
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "factors",
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.factors = "nfconv2",
reference.cat = FALSE, conv2.numsplits = 2,
AllLearner.real1 = 10)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "AllLearner", list(int1 = 2, real1 = 10), list(),
fnconv1 = list(reference.cat = FALSE),
nfconv2 = list(conv2.numsplits = 2))
pvs = list(selected.learner = "AllLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.before.impute = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "factors",
automlr.convert.factors = FALSE,
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.factors = "nfconv2",
conv2.numsplits = 2,
automlr.wimputing.factors = "factimputer1", fimp.const = "NAx",
AllLearner.real1 = 10, AllLearner.int1 = 6)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "AllLearner", list(int1 = 6, real1 = 10), list(),
nfconv2 = list(conv2.numsplits = 2),
factimputer1 = list(fimp.const = "NAx"))
pvs = list(selected.learner = "AllLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.before.impute = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "ordered",
automlr.convert.factors = TRUE,
automlr.convert.factors.to = "ordered",
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.ordered = "noconv2",
automlr.wconverting.factors.to.ordered = "foconv1",
automlr.wimputing.ordered = "ordimputer1", oimp.const = "NAx",
AllLearner.real1 = 10, AllLearner.int1 = 6)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "AllLearner", list(int1 = 6, real1 = 10), list(),
foconv1 = list(), noconv2 = list(),
ordimputer1 = list(oimp.const = "NAx"))
pvs = list(selected.learner = "AllLearner",
automlr.impute = FALSE, automlr.convert = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "ordered",
automlr.convert.factors = TRUE,
automlr.convert.factors.to = "ordered",
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.ordered = "noconv2",
automlr.wconverting.factors.to.ordered = "foconv1",
AllLearner.real1 = 10, AllLearner.int1.AMLRFIX2 = 12)
checkLearnerBehaviour(l, MissingsNumericsFactorsTask,
pvs, "AllLearner", list(int1 = 12, real1 = 10), list(),
foconv1 = list(), noconv2 = list())
})
# automlr.impute and automlr.wimputing are present in the right cases, and cause the right behaviour.
test_that("factors are present whenever missing indicators are introduced", {
# all learners and most removers
expect_warning(l <- buildLearners(list(
NumericsLearner, FactorsLearner, OrderedsLearner,
MissingsNumericsLearner, MissingsFactorsLearner, MissingsFactorsNumericsLearner,
FactorsNumericsLearner, AllLearner,
pWW(nimp1), pWW(nimp2), pWW(fimp1), pWW(fimp2), pWW(oimp1), pWW(oimp2),
pWW(fnconv1), pWW(onconv1), pWW(onconv2),
pWW(nfconv1), pWW(nfconv2), pWW(ofconv1), pWW(ofconv2),
pWW(noconv1), pWW(noconv2), pWW(foconv1), pWW(foconv2)), MissingsNumericsTask, verbosity = 6),
"different \\(but feasible\\) type 'cat' listed|has parameters .* not mentioned in search space", all = TRUE)
pvs = list(selected.learner = "NumericsLearner", automlr.missing.indicators = TRUE, automlr.convert.before.impute = TRUE,
automlr.wimputing.numerics = "numimputer2", multiplier = 1,
reference.cat = FALSE, NumericsLearner.int1 = 9)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "NumericsLearner", list(int1 = 9), list(),
fnconv1 = list(reference.cat = FALSE),
numimputer2 = list(multiplier = 1))
pvs = list(selected.learner = "NumericsLearner", automlr.missing.indicators = TRUE,
automlr.convert.before.impute = TRUE,
automlr.wimputing.numerics = "numimputer2", multiplier = 1,
reference.cat = FALSE, NumericsLearner.int1 = 9)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "NumericsLearner", list(int1 = 9), list(),
fnconv1 = list(reference.cat = FALSE),
numimputer2 = list(multiplier = 1))
pvs = list(selected.learner = "NumericsLearner", automlr.missing.indicators = FALSE,
automlr.wimputing.numerics = "numimputer2", multiplier = 1,
NumericsLearner.int1 = 9)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "NumericsLearner", list(int1 = 9), list(),
numimputer2 = list(multiplier = 1))
# The learner that can deal with missings and factors adds parameter automlr.impute. The wimputing / wconverting parameter is only present if the corresponding conversion happens.
pvs = list(selected.learner = "MissingsFactorsNumericsLearner",
automlr.convert.factors = TRUE, automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.numerics = FALSE,
automlr.missing.indicators = TRUE, automlr.convert.before.impute = FALSE,
automlr.wimputing.numerics = "numimputer1",
reference.cat = FALSE)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "MissingsFactorsNumericsLearner", list(), list(),
numimputer1 = list(),
fnconv1 = list(reference.cat = FALSE))
pvs = list(selected.learner = "MissingsFactorsNumericsLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.numerics = FALSE,
automlr.missing.indicators = FALSE,
automlr.wimputing.numerics = "numimputer1")
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "MissingsFactorsNumericsLearner", list(), list(),
numimputer1 = list())
pvs = list(selected.learner = "MissingsFactorsNumericsLearner",
automlr.convert.factors = TRUE, automlr.impute = FALSE, automlr.convert = TRUE,
automlr.convert.numerics = TRUE,
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.factors = "nfconv2",
reference.cat = FALSE, conv2.numsplits = 2,
MissingsFactorsNumericsLearner.int1 = 9)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "MissingsFactorsNumericsLearner", list(int1 = 9, real1 = 10), list(),
fnconv1 = list(reference.cat = FALSE),
nfconv2 = list(conv2.numsplits = 2))
pvs = list(selected.learner = "MissingsFactorsNumericsLearner",
automlr.impute = FALSE, automlr.convert = TRUE,
automlr.convert.numerics = TRUE,
automlr.missing.indicators = FALSE,
automlr.wconverting.numerics.to.factors = "nfconv2",
conv2.numsplits = 2,
MissingsFactorsNumericsLearner.int1 = 9)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "MissingsFactorsNumericsLearner", list(int1 = 9, real1 = 10), list(),
nfconv2 = list(conv2.numsplits = 2))
pvs = list(selected.learner = "MissingsFactorsNumericsLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.before.impute = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.factors = FALSE,
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.factors = "nfconv2",
conv2.numsplits = 2,
automlr.wimputing.factors = "factimputer1", fimp.const = "NAx",
MissingsFactorsNumericsLearner.bool1 = TRUE)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "MissingsFactorsNumericsLearner", list(bool1 = TRUE, real1 = 10), list(),
nfconv2 = list(conv2.numsplits = 2),
factimputer1 = list(fimp.const = "NAx"))
pvs = list(selected.learner = "MissingsFactorsNumericsLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.before.impute = TRUE,
automlr.convert.numerics = TRUE,
automlr.missing.indicators = FALSE,
automlr.wconverting.numerics.to.factors = "nfconv2",
conv2.numsplits = 2,
automlr.wimputing.factors = "factimputer1", fimp.const = "NAx",
MissingsFactorsNumericsLearner.bool1 = TRUE)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "MissingsFactorsNumericsLearner", list(bool1 = TRUE, real1 = 10), list(),
nfconv2 = list(conv2.numsplits = 2),
factimputer1 = list(fimp.const = "NAx"))
# AllLearner behaves essentially the same as MissingsFactorsNumericsLearner for the given task.
pvs = list(selected.learner = "AllLearner",
automlr.convert.factors = TRUE, automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.factors.to = "numerics",
automlr.convert.numerics = FALSE,
automlr.missing.indicators = TRUE, automlr.convert.before.impute = FALSE,
automlr.wimputing.numerics = "numimputer1",
reference.cat = FALSE,
AllLearner.int1 = 9, AllLearner.real1 = 8)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(bool1 = TRUE, int1 = 9, real1 = 8), list(),
numimputer1 = list(),
fnconv1 = list(reference.cat = FALSE))
pvs = list(selected.learner = "AllLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.numerics = FALSE,
automlr.missing.indicators = FALSE,
automlr.wimputing.numerics = "numimputer1",
AllLearner.int1 = 9, AllLearner.real1 = 8)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(bool1 = TRUE, int1 = 9, real1 = 8), list(),
numimputer1 = list())
pvs = list(selected.learner = "AllLearner",
automlr.convert.factors = TRUE, automlr.impute = FALSE, automlr.convert = TRUE,
automlr.convert.factors.to = "numerics",
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "factors",
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.factors = "nfconv2",
reference.cat = FALSE, conv2.numsplits = 2,
AllLearner.real1 = 10)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(int1 = 2, real1 = 10), list(),
fnconv1 = list(reference.cat = FALSE),
nfconv2 = list(conv2.numsplits = 2))
pvs = list(selected.learner = "AllLearner",
automlr.impute = FALSE, automlr.convert = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "factors",
automlr.missing.indicators = FALSE,
automlr.wconverting.numerics.to.factors = "nfconv2",
conv2.numsplits = 2,
AllLearner.real1 = 10)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(int1 = 2, real1 = 10), list(),
nfconv2 = list(conv2.numsplits = 2))
pvs = list(selected.learner = "AllLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.before.impute = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "factors",
automlr.convert.factors = FALSE,
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.factors = "nfconv2",
conv2.numsplits = 2,
automlr.wimputing.factors = "factimputer1", fimp.const = "NAx",
AllLearner.real1 = 10, AllLearner.int1 = 6)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(int1 = 6, real1 = 10), list(),
nfconv2 = list(conv2.numsplits = 2),
factimputer1 = list(fimp.const = "NAx"))
pvs = list(selected.learner = "AllLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.before.impute = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "factors",
automlr.missing.indicators = FALSE,
automlr.wconverting.numerics.to.factors = "nfconv2",
conv2.numsplits = 2,
automlr.wimputing.factors = "factimputer1", fimp.const = "NAx",
AllLearner.real1 = 10, AllLearner.int1 = 6)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(int1 = 6, real1 = 10), list(),
nfconv2 = list(conv2.numsplits = 2),
factimputer1 = list(fimp.const = "NAx"))
pvs = list(selected.learner = "AllLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.before.impute = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "ordered",
automlr.convert.factors = TRUE,
automlr.convert.factors.to = "ordered",
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.ordered = "noconv2",
automlr.wconverting.factors.to.ordered = "foconv1",
automlr.wimputing.ordered = "ordimputer1", oimp.const = "NAx",
AllLearner.real1 = 10, AllLearner.int1 = 6)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(int1 = 6, real1 = 10), list(),
foconv1 = list(), noconv2 = list(),
ordimputer1 = list(oimp.const = "NAx"))
pvs = list(selected.learner = "AllLearner",
automlr.impute = TRUE, automlr.convert = TRUE,
automlr.convert.before.impute = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "ordered",
automlr.missing.indicators = FALSE,
automlr.wconverting.numerics.to.ordered = "noconv2",
automlr.wimputing.ordered = "ordimputer1", oimp.const = "NAx",
AllLearner.real1 = 10, AllLearner.int1 = 6)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(int1 = 6, real1 = 10), list(),
noconv2 = list(),
ordimputer1 = list(oimp.const = "NAx"))
pvs = list(selected.learner = "AllLearner",
automlr.impute = FALSE, automlr.convert = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "ordered",
automlr.convert.factors = TRUE,
automlr.convert.factors.to = "ordered",
automlr.missing.indicators = TRUE,
automlr.wconverting.numerics.to.ordered = "noconv2",
automlr.wconverting.factors.to.ordered = "foconv1",
AllLearner.real1 = 10, AllLearner.int1.AMLRFIX2 = 12)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(int1 = 12, real1 = 10), list(),
foconv1 = list(), noconv2 = list())
pvs = list(selected.learner = "AllLearner",
automlr.impute = FALSE, automlr.convert = TRUE,
automlr.convert.numerics = TRUE,
automlr.convert.numerics.to = "ordered",
automlr.missing.indicators = FALSE,
automlr.wconverting.numerics.to.ordered = "noconv2",
AllLearner.real1 = 10, AllLearner.int1.AMLRFIX2 = 12)
checkLearnerBehaviour(l, MissingsNumericsTask,
pvs, "AllLearner", list(int1 = 12, real1 = 10), list(),
noconv2 = list())
})
test_that("targettype pseudoparameter is set as expected", {
nlr = NumericsLearner
nlr$searchspace = list(
sp("int1", "int", c(0, 0),
req = quote(automlr.targettype == "multiclass")),
sp("int1.AMLRFIX1", "int", c(1, 1),
req = quote(automlr.targettype == "twoclass")))
l = blt(list(nlr), NumericsTask)
checkLearnerBehaviour(l, NumericsTask, list(), "NumericsLearner",
list(int1 = 1), list())
tdat = getTaskData(NumericsTask)
tdat$fac.1 = factor(c("a", "b", "c", "b"))
NTask2 = makeClassifTask("NumericsTask2", tdat, "fac.1")
l = blt(list(nlr), NTask2)
checkLearnerBehaviour(l, NTask2, list(), "NumericsLearner",
list(int1 = 0), list())
})
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.