tests/testthat/test_local_runTaskFlow.R

In openml/r: Open Machine Learning and Open Data Platform

# skip("runTaskFlow skipped")
# 
# test_that("runTaskFlow", {
#   with_test_cache({
#     checkRun = function(res) {
#       expect_s3_class(res, "OMLMlrRun")
#       expect_equal(length(res), 3L)
#       expect_s3_class(res$run$predictions, "data.frame")
#       expect_s3_class(res$bmr, "BenchmarkResult")
#       expect_s3_class(res$flow, "OMLFlow")
#     }
#     getOMLRunEvaluations = function(run, measures, aggr = TRUE) {
#       evals = run$output.data$evaluations
#       evals = evals[evals$name %in% measures, ]
#       if (aggr) return(evals[is.na(evals$fold) & is.na(evals[, "repeat"]), ]) else
#         return(evals[!is.na(evals$fold) & !is.na(evals[, "repeat"]), ])
#     }
# 
#     task = getOMLTask(59)
# 
#     # do the same again and use a learner with parameter values
#     lrn = makeFilterWrapper(makeLearner("classif.randomForest", mtry = 2, ntree = 5), fw.method = "variance", fw.perc = 0.5)
#     run = getOMLRun(26936)
# 
#     flow = getOMLFlow(run$flow.id)
#     res2 = runTaskFlow(task, flow, par.list = getOMLRunParList(run), seed = getOMLSeedParList(run))
#     checkRun(res2)
#     par.list = convertOMLRunParListToList(getOMLRunParList(run))
#     res3 = runTaskFlow(task, flow, par.list = par.list, seed = 1)
#     checkRun(res3)
#     names(par.list)[1] = "renamed"
#     expect_error(runTaskFlow(task, flow, par.list = par.list, seed = 1))
# 
#     # compare with acc value computed from run
#     acc = getOMLRunEvaluations(run, "predictive_accuracy", aggr = FALSE)$value
#     acc2 = getBMRPerformances(res2$bmr)[[1]][[1]][["acc"]]
#     expect_equal(round(acc, 5), round(acc2, 5))
# 
#     # check if changing the seed yields different results (for RF it should)
#     res3 = runTaskFlow(task, flow, par.list = getOMLRunParList(run), seed = 123456)
#     checkRun(res3)
#     acc3 = getBMRPerformances(res3$bmr)[[1]][[1]][["acc"]]
#     expect_false(isTRUE(all.equal(round(acc, 5), round(acc3, 5))))
# 
#     # check if passing parameters as list works
#     lrn = makeLearner("classif.randomForest", mtry = 2)
#     task = getOMLTask(task.id = 59L)
#     run.mlr = runTaskMlr(task, lrn)
#     run.flow = runTaskFlow(task, convertMlrLearnerToOMLFlow(lrn), par.list = getHyperPars(lrn))
#     checkRun(run.flow)
# 
#     expect_equal(getBMRPerformances(run.mlr$bmr, as.df = TRUE)$acc,
#       getBMRPerformances(run.flow$bmr, as.df = TRUE)$acc)
#   })
# })