mlr (version 2.7)

friedmanTestBMR: Perform overall Friedman test for a BenchmarkResult.

Description

Performs a friedman.test for a selected measure. The null hypothesis is that apart from an effect of the different [tasks], the location parameter (aggregated performance-measure) is the same for each learner.

Usage

friedmanTestBMR(bmr, measure = NULL, aggregation = "default")

Value

A list of class htest. See friedman.test for details.

See Also

Other benchmark: BenchmarkResult, benchmark, convertBMRToRankMatrix, friedmanPostHocTestBMR, generateCritDifferencesData, getBMRAggrPerformances, getBMRFeatSelResults, getBMRFilteredFeatures, getBMRLearnerIds, getBMRLearners, getBMRMeasureIds, getBMRMeasures, getBMRPerformances, getBMRPredictions, getBMRTaskIds, getBMRTuneResults, plotBMRBoxplots, plotBMRRanksAsBarChart, plotBMRSummary, plotCritDifferences

Examples

lrns = list(makeLearner("classif.nnet"), makeLearner("classif.rpart"))
tasks = list(iris.task, sonar.task)
rdesc = makeResampleDesc("CV", iters = 2L)
res = benchmark(lrns, tasks, rdesc, acc)
friedmanTestBMR(res)