CustomIterator

class lightautoml.validation.base.CustomIterator(train, iterator)[source]

Iterator that uses function to create folds indexes.

Usefull for example - classic timeseries splits.

__init__(train, iterator)[source]

Create iterator.

Parameters:

train (LAMLDataset) – Dataset of train data.
iterator (Iterable[Tuple[Sequence, Sequence]]) – Callable(dataset) -> Iterator of train/valid indexes.

get_validation_data()[source]

Simple return train dataset.

convert_to_holdout_iterator()[source]

Convert iterator to hold-out-iterator.

Use first train/valid split for HoldoutIterator creation.