MarkovSwitchingVARResult._from_fit#

classmethod MarkovSwitchingVARResult._from_fit(fit, model)[source]#

Assemble the public result, building one stable view per regime.

Each regime view carries the regime’s own conditional-mean residuals rescaled by the square root of its smoothed weight (normalized to the sample size), so that the view’s resid.T @ resid / nobs is the regime’s probability-weighted residual covariance.

Parameters:

  • fit (_VectorMarkovSwitchingFit) – The EM output: transition matrix, per-regime coefficient blocks and covariances, the three posteriors, the mixture fitted values and residuals.

  • model (_MarkovSwitchingVectorAutoRegressionModel[MarkovSwitchingVARResult]) – The model that produced it, read for the sample, the specification and the ordering convention.

Return type:

MarkovSwitchingVARResult

Returns:

A populated result whose regimes are identity-stable.

Example

>>> import numpy as np
>>> rng = np.random.default_rng(0)
>>> s = np.repeat([0, 1, 0, 1], 100)
>>> y = np.array([[-1.0, -0.5], [1.5, 1.0]])[s] + 0.4 * rng.standard_normal((400, 2))
>>> model = MarkovSwitchingVAR(y, order=1, n_regimes=2)
>>> fit = model._fit_markov(max_iter=500, tol=1e-6, n_init=10, screen_iter=15, seed=0)
>>> res = MarkovSwitchingVARResult._from_fit(fit, model)
>>> view = res.regimes[0]
>>> weighted = view.resid.T @ view.resid / res.nobs
>>> round(view.nobs), bool(np.allclose(weighted, view.sigma_u, atol=1e-6))
(199, True)