- BREAKING CHANGE: The
kknn
package was removed from CRAN. Thekknn
tuning spaces are removed from the package. - feat: Print with
cli
package.
- compatibility: mlr3learners 0.9.0
- compatibility: Work with new mlr3tuning version 1.0.0
- fix: Reduce levels of
respect.unordered.factors
of search spaceranger.rbv1
to"ignore"
and"order"
. - compatibility: Work with new paradox version 1.0.0
- fix: The source of the
rbv2
search spaces was wrong. The source is Binder, Pfisterer, and Bischl (2020). Thanks to @markus-schaffer. - feat: Add
rbv1
search spaces from the Kühn (2018) paper.
- fix: Add missing
...
parameter tord_info.TuningSpace()
.
- docs: Remove deprecated
method
argument ofmlr3tuning::tune()
. - docs: Update resources.
- fix: Extra
paradox::TuneToken
inlts()
were not passed to learners created with$get_learner()
. - docs: Add
lts()
return.
- docs: Add
mlr_tuning_spaces
prefix to aliases.
- docs: Add glmnet description.
- feat: Pass
paradox::TuneToken
tolts()
to add, remove or overwrite parameters in tuning spaces and learners. - fix: Remove debugging from
lts()
.
- feat: Add a
as.data.table.TuningSpace()
function. - feat:
TuningSpace
objects have the optional field$label
now. - feat: New
$help()
method which opens the manual page of aTuningSpace
. - feat: Add search space for
glmnet
andkknn
to default collection. - feat: New
as_search_space()
function to create search spaces fromTuningSpace
objects.
- fix: The
subsample
hyperparameter is tuned on a logarithmic scale now. The lower bound ofalpha
is reduced from1e-4
to1e-3
. The tuning range of thelambda
hyperparameter was 0.1 to 1. From now on,lambda
is tuned from1e-3
to1e3
on a logarithmic scale.
- refactor: update citations.
- feat: Add
mtry.ratio
hyperparameter to tuning spaces of the ranger learner. - feat: Add
$print()
method toTuningSpace
objects.
- First release of the tuning spaces package.