DecompositionFilterResult._method_notes#

DecompositionFilterResult._method_notes()[source]#

What the method’s definition of the trend implies for reading the cycle.

The penalized filters report the provisional row count and, for Hodrick-Prescott, Hamilton’s (2018) critique; Hamilton’s filter states its one-sidedness and the rows it discards; Beveridge-Nelson, the fall-through branch, explains why its cycle is small and volatile.

Return type:

tuple[str, ...]

Returns:

One or two note strings for the summary’s closing block.

Example

>>> rng = np.random.default_rng(0)
>>> y = np.cumsum(rng.standard_normal(200))
>>> notes = HamiltonFilter().filter(y)._method_notes()
>>> len(notes), notes[0][:47]
(2, 'One-sided by construction: the cycle at t is wh')
>>> ButterworthFilter(period=32).filter(y)._method_notes()[0][:60]
'Two-sided Butterworth sine filter of Gomez (2001); the 64 ro'