Michel Lang
As a proof of concept. Also implement `fix_factor_levels.Matrix()`.
C.f. #432.
They have been removed for now to simplify debugging and refactoring. But they are definitely useful, they should be re-introduced.
requires new mlr3
I.e., mlr3extralearners and mlr3proba is not explained. Not sure about mlr3verse.
Something is already in the mlr3pipelines attic.
As suggested in https://github.com/mlr-org/mlr3tuning/issues/14.
For an example, see https://www.openml.org/api/v1/json/data/3. The `upload_date` is returned in format "%Y-%m-%dT%H:%M:%S", the `processing_date` as "%Y-%m-%d %H:%M:%S" (space instead of 'T').
https://www.openml.org/api/v1/json/data/40984 reports three ignore attributes ("region.centroid.row", "region.centroid.col", "region.centroid.count"). https://www.openml.org/api/v1/json/data/features/40984 only one ("region.centroid.col"). Which one is right? Also note that this is a data set used in CC18.