HierarchicalBVARResult._summary_table#

HierarchicalBVARResult._summary_table()[source]#

Build the structured summary.

Return type:

SummaryTable

Returns:

One row per hyperparameter with its mode and, under the full method, its posterior mean and 68% interval; the metadata carries the method, the draw count, the dimensions and the log marginal likelihood at the mode; the notes state the hyperpriors and the exploration, the acceptance rate (full method), what the reported evidence conditions on, and the stability share.

Example

>>> import numpy as np
>>> rng = np.random.default_rng(0)
>>> y = 5.0 + rng.standard_normal((150, 2))
>>> empirical = HierarchicalBVAR(y, order=1).fit(
...     method="empirical", n_draws=50, seed=0
... )
>>> table = empirical._summary_table()
>>> table.title, table.columns[:2], table.metadata[1]
('Hierarchical BVAR(1) Results', ('hyperparameter', 'mode'), ('Method', 'empirical'))
>>> table.rows[0][2:], len(table.notes)
(('-', '-'), 3)
>>> full = HierarchicalBVAR(y, order=1).fit(
...     method="full", n_draws=100, n_burn=100, seed=0
... )
>>> len(full._summary_table().notes), full._summary_table().notes[1][:26]
(4, 'Metropolis acceptance rate')