NSSDNResult.compare#

NSSDNResult.compare(*others, criterion='aic')[source]#

Rank this result and others by one criterion, best first.

Parameters:

  • others (_ComparisonMixin) – Results fitted on the same effective sample (the same nobs); any mixture of result types that carry the three attributes.

  • criterion (str) – "aic", "bic" or "hqic".

Return type:

SummaryTable

Returns:

A table with one row per result – the label, llf, the parameter count under k, the criterion and its difference from the best (zero for the first row).

Raises:

SpecificationError – If the criterion is unknown or the results do not share nobs.

Example

>>> import numpy as np
>>> from ns_sdn.spectral_decomposition import NSSDN, Stage
>>> y = np.sin(0.3 * np.arange(80.0))
>>> one = NSSDN(y, k=1, state_dim=4, schedule=(Stage(1e-2, 5),), seed=0).fit()
>>> two = NSSDN(y, k=2, state_dim=4, schedule=(Stage(1e-2, 5),), seed=0).fit()
>>> table = one.compare(two, criterion="bic")
>>> table.columns, table.rows[0][-1]
(('model', 'llf', 'k', 'BIC', 'dBIC'), '0.000')
>>> one.compare(two, criterion="r2")
Traceback (most recent call last):
    ...
ns_sdn.exceptions.SpecificationError: criterion must be 'aic', 'bic' or 'hqic'; got ...
On this page