smooth.ADAM.holdout_data

property ADAM.holdout_data: ndarray[tuple[int, ...], dtype[_ScalarType_co]] | None

$holdout).

If holdout=True was specified during fitting, returns the portion of data withheld for validation. Otherwise returns None.

Returns:

1D array of holdout observations, or None if no holdout was used.

Return type:

Optional[NDArray]

Raises:

ValueError – If the model has not been fitted yet.

Examples

>>> model = ADAM(model="AAN", holdout=True, h=12)
>>> model.fit(y)
>>> if model.holdout_data is not None:
...     print(f"Holdout size: {len(model.holdout_data)}")

Type:

Holdout validation data (R