luz.scorers module¶

class CrossValidation(num_folds, val_fraction=None, fold_seed=None, shuffle=True)¶

Object which scores a learning algorithm using cross validation.

Parameters

num_folds (int) – Number of cross validation folds.
val_fraction (Optional[int]) – Fraction of data to use as a validation set, by default None.
fold_seed (Optional[int]) – Seed for random fold split, by default None.
shuffle (Optional[bool]) – If True, shuffle dataset before splitting into folds; by default True.

score(learner, dataset, device='cpu')¶

Learn a model and estimate its future performance using cross validation.

Parameters

learner (Learner) – Learning algorithm to be scored.
dataset (Dataset) – Dataset to use for scoring.
device (Union[str, device, None]) – Device to use for scoring, by default “cpu”.

Returns

Learned model and cross-validation score.

Return type

luz.Score

class Holdout(test_fraction, val_fraction=None)¶

Object which scores a learning algorithm using the holdout method.

Parameters

test_fraction (float) – Fraction of data to use as a test set for scoring.
val_fraction (Optional[float]) – Fraction of data to use as a validation set, by default None.

score(learner, dataset, device='cpu')¶

Learn a model and estimate its future performance using the holdout method.

Parameters

learner (Learner) – Learning algorithm to be scored.
dataset (Dataset) – Dataset to use for scoring.
device (Union[str, device, None]) – Device to use for scoring, by default “cpu”.

Returns

Learned model and holdout score.

Return type

luz.Score

class Score(model, score)¶

Bases: tuple

Create new instance of Score(model, score)

class Scorer¶

Bases: abc.ABC

abstract score(learner, dataset, device)¶