Version 2.3
mlr 2.3:
CRAN release: 2015-02-04
- resample now returns an object of class ResampleResult (downward compatible) to allow for a print method.
- resampling on features now supported for an arbitrary number of factor features
- mlr supports ViperCharts plots now
- ROC plot via ROCR can now be created automatically, before you had to call asROCRPrediction, then construct the plots via ROCR your self. See plotROCRCurves
- all mlr measures now have slots “name” and “note”
- exported a few very simple “getters” for tasks, see below
- in makeLearner a probability predict.threshold can be set for classifiers, also see setPredictThreshold
- in the control objects for tuning and feature selection, the user can now enable threshold tuning
- in the control objects for tuning and feature selection, the user can now define his own logging function
- default console logging for tuneParams and selectFeatures is more informative, it displays time and memory info
- updated some properties of some learners
- Default arguments of classif.bartMachine, classif.randomForestSRC, regr.randomForestSRC and sur.randomForestSRC have been changed to allow missing data support with default settings.
- externalized measure functions to be used on vectors.
- some minor bug fixes
- required basic learner packages are not loaded into the global namespace anymore, requireNamespace is used internally instead. this ensures less name clashes and name shadowing
- resample passes dot arguments to the learner hyperpars
- new option “on.par.out.of.bounds” to disable out-of-bound checks for model parameters
- measures were slightly internally changed. they expose more properties (check ?Measure) and some now unnecessary object slots were removed
- classif.lda and classif.qda now have hyperpar “predict.method”
- filterFeatures and makeFilterWrapper gain an argument for mandatory features
- plotLearnerPrediction has new option “err.size”
- classif.plsDA and cluster.DBscan for now removed because of problems with the underlying learning algorithm
- new aggregation test.join
- the following models now can handle factors and ordereds by extra dummy or int encoding: classif.glmnet, regr.glmnet, surv.glmnet, surv.cvglmnet, surv.penalized, surv.optimCoxBoostPenalty, surv.glmboost, surv.CoxBoost