StabilityWarning

exception cultivars.exceptions.StabilityWarning

Bases: UserWarning

The warning category for a fitted model that is non-stationary.

A fit that converged to a companion matrix with a root on or outside the unit circle is still a fit – the estimates are reported, and the result’s is_stable and stability_check() say what it found – but forecasts, impulse responses and spectral densities from it are not the objects their names promise. This is the category a layer uses when it chooses to warn about that rather than refuse; a UserWarning rather than a CultivarsError because nothing has failed. Filter it with warnings.filterwarnings() when the non-stationarity is the point of the exercise. Results also report instability in their summary notes, which are not affected by the filter.

See also

• NumericalError – raised, not warned, when a computation requires stationarity and does not have it.

Example

The category can be silenced or promoted on its own, leaving other warnings alone:

>>> import warnings
>>> with warnings.catch_warnings(record=True) as caught:
...     warnings.simplefilter("always")
...     warnings.warn("root on the unit circle", StabilityWarning)
>>> issubclass(caught[0].category, UserWarning), caught[0].category.__name__
(True, 'StabilityWarning')
>>> with warnings.catch_warnings(record=True) as caught:
...     warnings.simplefilter("always")
...     warnings.filterwarnings("ignore", category=StabilityWarning)
...     warnings.warn("root on the unit circle", StabilityWarning)
...     warnings.warn("something else", UserWarning)
>>> [w.category.__name__ for w in caught]
['UserWarning']