Due to the high amount of packages in the mlr3 ecosystem, it is hard to keep up with the latest changes across all packages. This posts gives an overview by listing the recent release notes of mlr3 packages from the last quarter. Note that only CRAN packages are listed here and the sort order is alphabetically.
Interval: 2021-10-01 – 2021-12-31
mlr3 0.13.0 – https://github.com/mlr-org/mlr3
Description: Machine Learning in R – Next Generation
- Learners which are capable of resuming/continuing (e.g.,
nroundsupdated) can now optionally store a stack of trained learners to be used to hotstart their training. Note that this feature is still somewhat experimental. See
- New measures to score similarity of selected feature sets:
sim.jaccard(Jaccard Index) and
sim.phi(Phi coefficient) (#690).
predict_newdata()now also supports
- New function
install_pkgs()to install required packages. This generic works for all objects with a
packagesfield as well as
- New learner
$set_levels()to control how data with factor columns is returned, independent of the used
- Measures now return
NAif prerequisite are not met (#699). This allows to conveniently score your experiments with multiple measures having different requirements.
- Feature names may no longer contain the special character
mlr3benchmark 0.1.3 – https://github.com/mlr-org/mlr3benchmark
Description: Analysis and Visualisation of Benchmark Experiments
- Fix README.
- Fix for PMCMRplus.
mlr3db 0.4.2 – https://github.com/mlr-org/mlr3db
Description: Data Base Backend for ‘mlr3’
- Compatibility fixes with new duckdb version.
mlr3learners 0.5.1- – https://github.com/mlr-org/mlr3learners
Description: Recommended learners for mlr3
eval_metric()is now explicitly set for xgboost learners to silence a deprecation warning.
- Improved how the added hyperparameter
mtry.ratiois converted to
mtryto simplify tuning.
- Multiple updates to hyperparameter sets.
mlr3pipelines 0.4.0 – https://github.com/mlr-org/mlr3pipelines
Description: Preprocessing Operators and Pipelines for ‘mlr3’
- New operator
%>>!%that modifies Graphs in-place.
- New methods
Graph$chain()as alternatives for
- New methods
ppls()which create lists of PipeOps/Graphs and can be seen as “plural” forms of
po()S3-method for PipeOp class that clones a PipeOp object and optionally modifies its attributes.
Graph$add_pipeop()now clones the PipeOp being added.
- Documentation: Clarified documentation about cloning of input arguments in several places.
- Performance enhancements for Graph concatenation.
- More informative error outputs.
- New attribute graph_model
GraphLearnerclass, which gets the trained graph.
as_learner()S3-method for PipeOp class that turns wraps a PipeOp in a Graph and turns that into a Learner.
- Changed PipeOps:
PipeOpHistBin: renamed ‘bins’ Param to ‘breaks’
PipeOpImputeHist: fix handling of integer features spanning the entire represented integer range
PipeOpImputeOOR: fix handling of integer features spanning the entire represented integer range
PipeOpProxy: Avoid unnecessary clone
PipeOpScale: Performance improvement
mlr3proba 0.4.2 – https://github.com/mlr-org/mlr3proba
Description: Probabilistic Supervised Learning for ‘mlr3’
- Patch for linux.
mlr3spatial 0.1.0 https://github.com/mlr-org/mlr3spatial
Description: Support for Spatial Objects Within the ‘mlr3’ Ecosystem
- Initial release.
mlr3tuningspaces 0.0.1 – https://github.com/mlr-org/mlr3tuningspaces
Description: Search Spaces for Hyperparameter Tuning
- Initial release.
mlr3viz 0.5.7 – https://github.com/mlr-org/mlr3viz
Description: Visualizations for ‘mlr3’
- Compatibility fix for testthat.