RegimeSwitchingLinearSSM.smooth#
- RegimeSwitchingLinearSSM.smooth(y)[source]#
Kim’s backward pass over the regime probabilities.
Runs the forward filter, then the discrete backward recursion the package already owns for Hamilton-filter models, applied to the collapse-approximate filtered probabilities. Smoothing the continuous state of a switching model requires a second collapse approximation on the backward pass and is deliberately not offered; the regime chronology is what smoothing is for here.
- Parameters:
y (
ArrayLike) – Data,(n, p);numpy.nanrows are missing.- Return type:
_KimSmootherResult- Returns:
The
_KimSmootherResultover regimes.