SparkFoldsIterator

class sparklightautoml.validation.iterators.SparkFoldsIterator(train, n_folds=None)

Bases: SparkBaseTrainValidIterator

Classic cv iterator.

Folds should be defined in Reader, based on cross validation method.

__init__(train, n_folds=None)

Creates iterator.

Parameters:

• train (SparkDataset) – Dataset for folding.

• n_folds (Optional[int]) – Number of folds.

convert_to_holdout_iterator()

Convert iterator to hold-out-iterator.

Fold 0 is used for validation, everything else is used for training.

Return type:

SparkHoldoutIterator

Returns:

new hold-out-iterator.