get_numpy_iterator

lightautoml.validation.np_iterators.get_numpy_iterator(train, valid=None, n_folds=None, iterator=None)

Get iterator for np/sparse dataset.

If valid is defined, other parameters are ignored. Else if iterator is defined n_folds is ignored.

Else if n_folds is defined iterator will be created by folds index. Else DummyIterator - (train, train) will be created.

Parameters:

• train (Union[CSRSparseDataset, NumpyDataset, PandasDataset]) – LAMLDataset to train.

• valid (Union[CSRSparseDataset, NumpyDataset, PandasDataset, None]) – Optional LAMLDataset for validate.

• n_folds (Optional[int]) – maximum number of folds to iterate. If None - iterate through all folds.

• iterator (Optional[Iterable[Tuple[Sequence, Sequence]]]) – Takes dataset as input and return an iterator of indexes of train/valid for train dataset.

Return type:

Union[FoldsIterator, HoldoutIterator, CustomIterator, DummyIterator]

Returns:

new train-validation iterator.