Machine Learning with mlr

The mlr package provides a unified interface for supervised and unsupervised machine learning in R, covering task creation, learner configuration, resampling, performance measurement, hyperparameter tuning, and visualization. It supports classification, regression, clustering, survival, and cost-sensitive tasks.

What’s covered#

• Setup – normalizeFeatures(), createDummyFeatures(), mergeSmallFactorLevels()
• Task creation – makeClassifTask(), makeRegrTask(), makeClusterTask(), makeSurvTask()
• Training & testing – makeLearner(), train(), predict(), performance(), listMeasures()
• Resampling – makeResampleDesc() for CV, LOO, RepCV, bootstrap, and holdout; resample()
• Refining performance – makeParamSet(), makeTuneControl*(), tuneParams()
• Configuration – configureMlr() for error handling and output verbosity
• Feature extraction – filterFeatures() with importance-based selection methods
• Visualization – plotThreshVsPerf(), plotROCCurves(), plotResiduals()
• Wrappers – makeDummyFeaturesWrapper(), makeImputeWrapper(), makePreprocWrapper()