skdatasets.utils.experiment.fetch_scores

skdatasets.utils.experiment.fetch_scores(*, storage: sacred.observers.base.RunObserver | str, ids: Optional[Sequence[int]] = None, dataset_names: Optional[Sequence[str]] = None, estimator_names: Optional[Sequence[str]] = None) → ScoresInfo

Fetch scores from Sacred experiments.

By default, it retrieves every experiment. The parameters ids, estimator_names and dataset_names can be used to restrict the number of experiments returned.

Parameters:

• storage (sacred.observers.RunObserver or str) – Where the experiments are stored. Either a Sacred observer, for example for a Mongo database, or the name of a directory, to use a file observer.

• ids (Sequence of int or None, default None) – If not None, return only experiments whose id is contained in the sequence.

• dataset_names (Sequence of str or None, default None) – If not None, return only experiments whose dataset names are contained in the sequence. The order of the names is also the one used for datasets when combining the results.

• estimator_names (Sequence of str or None, default None) – If not None, return only experiments whose estimator names are contained in the sequence. The order of the names is also the one used for estimators when combining the results.

Returns:

info – Class containing information about experiments scores.

Return type:

ScoresInfo

See also

run_experiments, fetch_scores