Glossary¶
This glossary aims to describe the concepts and either detail their corresponding API, or link to other relevant parts of the documentation which do so. By linking to glossary entries from the API Reference and User Guide, we may minimize redundancy and inconsistency.
Most of the concepts are available in the glossary of scikit-learn, but more specific sets of related terms are listed below:
General concepts¶
- ranking
- rankings
- partial ranking
- partial rankings
An order of preference of (some) discrete values defined over a finite set. Rankings may be complete (all values are ranked) or incomplete (only some values are ranked). Rankings may also be classified as with-ties (partial rankings) or without-ties (rankings) depending on if they present lack of preference information among some of the ranked values.
Class APIs and estimator types¶
- label ranker
- label rankers
A supervised predictor with a ranking defined over a finite set of discrete values.
Label rankers usually inherit from
base.LabelRankerMixin, which sets their _estimator_type attribute.A label ranker can be distinguised from other estimators with
is_label_ranker.- partial label ranker
- partial label rankers
A supervised predictor with a partial ranking defined over a finite set of discrete values.
Partial label rankers usually inherit from
base.PartialLabelRankerMixin, which sets their _estimator_type attribute.A partial label ranker can be distinguised from other estimators with
is_partial_label_ranker.A partial label ranker must implement: fit, predict and score.