cultivars.summary#

The reporting surface: the records every fitted result renders itself through.

A fitted result is a frozen record, and summary() does not format it; it returns a SummaryTable, a second record holding the structure of the report – a title, a block of scalar metadata, one coefficient table, and the closing notes that state what the numbers assume – which renders on demand as fixed-width text, as HTML in a notebook, or as a dataframe. InformationCriteria is the record behind every information_criteria property: the three penalized likelihoods,

\[\mathrm{AIC} = -2\ell + 2k, \qquad \mathrm{BIC} = -2\ell + k \log n, \qquad \mathrm{HQIC} = -2\ell + 2k \log\log n,\]

for a maximized log-likelihood \(\ell\), parameter count \(k\) and sample size \(n\), with the convention for \(k\) stated by the result that produced it.

Two commitments shape the surface. First, the two records are defined once, in _core, and this module is their public name: every summary() and information_criteria in the package returns an instance of the classes documented here, so their methods and fields are stable API, and the engine pages document the same objects only as internals. Second, nothing here is specific to a model family; a record that needs more structure than a title, metadata, one table and notes belongs beside the result that needs it.

Layout. SummaryTable with its to_text, to_html, to_pandas and to_polars renderers, and InformationCriteria with from_likelihood. Both live in _core._containers and are bound to this module so that annotations, pickles and the API reference all name them here.

Example

>>> import numpy as np
>>> from cultivars.summary import InformationCriteria, SummaryTable
>>> from cultivars.univariate.box_jenkins import ARIMA
>>> res = ARIMA(np.random.default_rng(0).standard_normal(120), order=(1, 0, 0)).fit()
>>> isinstance(res.summary(), SummaryTable)
True
>>> isinstance(res.information_criteria, InformationCriteria)
True
>>> SummaryTable.__module__, SummaryTable.__qualname__
('cultivars.summary', 'SummaryTable')

Classes#

InformationCriteria

Model-selection criteria computed from a fitted log-likelihood.

SummaryTable

A rendered-on-demand estimation summary.
On this page