Version 0.3#

scikit-lr has been rewritten on top of the current scikit-learn API.

Legend for changelogs

Major Feature something big that you could not do before.
Feature something that you could not do before.
Efficiency an existing feature now may not require as much computation or memory.
Enhancement a miscellaneous minor improvement.
Fix something that previously did not work as documented, or according to reasonable expectations, should now work.
API Change you will need to change your code to have the same effect in the future, or a feature will be removed in the future.

Version 0.3.0#

In development

The extension modules extend the internals of scikit-learn, so scikit-lr is built against a single minor release of scikit-learn, 1.9, and importing sklr with another one raises an ImportError. By Juan Carlos Alfaro Jiménez. #99
There are wheels for Linux (x86_64 and aarch64), macOS (x86_64 and arm64) and Windows (AMD64 and ARM64). By Juan Carlos Alfaro Jiménez. #99
The public API has type hints, which type checkers find in the installed package (PEP 561). By Juan Carlos Alfaro Jiménez. #99
scikit-lr requires Python 3.13 or later, and is built with meson-python. By Juan Carlos Alfaro Jiménez. #99

Feature show_versions() prints the versions of scikit-lr, of the scikit-learn that it was built against and of the dependencies, to include them in bug reports. By Juan Carlos Alfaro Jiménez. #99

Feature base.LabelRankerMixin and base.PartialLabelRankerMixin set the tags of the label rankers and of the partial label rankers, and their score method computes metrics.kendall_tau_score() and metrics.tau_x_score() on the predictions of the ranker. base.is_label_ranker() and base.is_partial_label_ranker() tell them apart from other estimators. By Juan Carlos Alfaro Jiménez. #99

Feature metrics.kendall_tau_score() and metrics.kendall_distance() compare label rankings, and metrics.tau_x_score() compares partial label rankings, averaging over the samples with optional sample weights. metrics.get_scorer() gets their scorers by name, to pass them to the scoring parameter of the model selection tools of scikit-learn, and metrics.get_scorer_names() lists the names. By Juan Carlos Alfaro Jiménez. #99

Feature utils.type_of_ranking() determines whether a target holds label rankings, partial label rankings or does not encode rankings, following the representation described in Ranking representation. By Juan Carlos Alfaro Jiménez. #99
Feature utils.check_ranking() validates the rankings of a target and converts them to a float64 array, rejecting rankings with ties and incomplete rankings unless they are allowed. By Juan Carlos Alfaro Jiménez. #99