mlr3 package updates – Q3/2021

[This article was first published on Machine Learning in R, and kindly contributed to R-bloggers]. (You can report issue about the content on this page here)
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.

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 post tries to tackle this issue by listing all release notes of the packages most recent releases in the last quarter. Note that only CRAN packages are listed here and the sort order is alphabetically.

Interval: 2021-07-01 – 2021-10-01

bbotk 0.4.0 – https://github.com/mlr-org/bbotk

Description: Black-Box Optimization Toolkit

  • Adds non dominated sorting with hypervolume contribution to Archive.
  • Allows empty search space and domain.
  • Extended TerminatorEvals with an additional hyperparameter k to define the budget depending on the dimension of the search space.
  • Adds bb_optimize() function for quick optimization.
  • Adds OptimizerIrace from irace package.

mlr3 0.12.0 – https://github.com/mlr-org/mlr3

Description: Machine Learning in R – Next Generation

  • New method to assign labels to columns in tasks: Task$label(). These will be used in visualizations in the future.
  • New method to add stratification variables: Task$add_strata().
  • New helper function partition() to split a task into a training and test set.
  • New standardized getter loglik() for class Learner.
  • New measures "aic" and "bic" to compute the Akaike Information Criterion or the Bayesian Information Criterion, respectively.
  • New Resampling method: ResamplingCustomCV. Creates a custom resampling split based on the levels of a user-provided factor variable.
  • New argument encapsulate for resample() and benchmark() to conveniently enable encapsulation and also set the fallback learner to the featureless learner. This is simply for convenience, configuring each learner individually is still possible and allows a more fine-grained control (#634, #642).
  • New field parallel_predict for Learner to enable parallel predictions via the future backend. This currently is only enabled while calling the $predict() or $predict_newdata methods and is disabled during resample() and benchmark() where you have other means to parallelize.
  • Deprecated public (and already documented as internal) field $data in ResampleResult and BenchmarkResult to simplify the API and avoid confusion. The converter as.data.table() can be used instead to access the internal data.
  • Measures now have formal hyperparameters. A popular example where this is required is the F1 score, now implemented with customizable beta.
  • Changed default of argument ordered in Task$data() from TRUE to FALSE.

mlr3cluster 0.1.2 – https://github.com/mlr-org/mlr3cluster

Description: Cluster Extension for ‘mlr3’

  • Add Hclust
  • test and doc hclust
  • Add within sum of squares measure
  • add doc wss
  • code factor adaptions

mlr3filters 0.4.2 – https://github.com/mlr-org/mlr3filters

Description: Filter Based Feature Selection for ‘mlr3’

  • Fixes an issue where argument nfeat was not passed down to {praznik} filters (#97)

mlr3learners 0.5.1 – https://github.com/mlr-org/mlr3learners

Description: Recommended Learners for ‘mlr3’

  • Improved how the added hyperparameter mtry.ratio is converted to mtry to simplify tuning.

mlr3pipelines 0.3.4 – https://github.com/mlr-org/mlr3pipelines

Description: Preprocessing Operators and Pipelines for ‘mlr3’

  • Stability: PipeOps don’t crash when they have python/reticulate hyperparameter values.
  • Documentation: Titles of PipeOp documentation articles reworked.

mlr3spatiotempcv 1.0.0 – https://github.com/mlr-org/mlr3spatiotempcv

Description: Spatiotemporal Resampling Methods for ‘mlr3’

Breaking

  • autoplot(): removed argument crs. The CRS is now inferred from the supplied Task. Setting a different CRS than the task might lead to spurious issues and the initial idea of changing the CRS for plotting to have proper axes labeling does not apply (anymore) (#144)

Features

  • Added autoplot() support for ResamplingCustomCV (#140)

Bug fixes

  • "spcv_block": Assert error if folds > 2 when selection = "checkerboard" (#150)
  • Fixed row duplication when creating TaskRegrST tasks from sf objects (#152)

Miscellaneous

  • Upgrade tests to {vdiffr} 1.0.0
  • Add {rgdal} to suggests and required it in "spcv_block" since it is required in {blockCV} >= 2.1.4 and {sf} >= 1.0

mlr3tuning 0.9.0 – https://github.com/mlr-org/mlr3tuning

Description: Tuning for ‘mlr3’

  • Adds AutoTuner$base_learner() method to extract the base learner from nested learner objects.
  • tune() supports multi-criteria tuning.
  • Allows empty search space.
  • Adds TunerIrace from irace package.
  • extract_inner_tuning_archives() helper function to extract inner tuning archives.
  • Removes ArchiveTuning$extended_archive() method. The mlr3::ResampleResults are joined automatically by as.data.table.TuningArchive() and extract_inner_tuning_archives().

mlr3verse 0.2.3 – https://github.com/mlr-org/mlr3verse

Description: Easily Install and Load the ‘mlr3’ Package Family

  • Updated reexports.

mlr3viz 0.5.6 – https://github.com/mlr-org/mlr3viz

Description: Visualizations for ‘mlr3’

  • Compatibility fix for mlr3tuning.
  • Fixed position of labels in barplot for PredictionClassif.

mlr3viz 0.5.5

  • Fixed another bug for ROC- and Precision-recall-curves (#79).

To leave a comment for the author, please follow the link and comment on their blog: Machine Learning in R.

R-bloggers.com offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.

Never miss an update!
Subscribe to R-bloggers to receive
e-mails with the latest R posts.
(You will not see this message again.)

Click here to close (This popup will not appear again)