statsmodels.tsa.kalmanf.kalmanfilter.KalmanFilter.loglike

classmethod KalmanFilter.loglike(params, arma_model, set_sigma2=True)[source]

The loglikelihood for an ARMA model using the Kalman Filter recursions.

Parameters:

params : array

The coefficients of the ARMA model, assumed to be in the order of trend variables and k exogenous coefficients, the p AR coefficients, then the q MA coefficients.

arma_model : statsmodels.tsa.arima.ARMA instance

A reference to the ARMA model instance.

set_sigma2 : bool, optional

True if arma_model.sigma2 should be set. Note that sigma2 will be computed in any case, but it will be discarded if set_sigma2 is False.

Notes

This works for both real valued and complex valued parameters. The complex values being used to compute the numerical derivative. If available will use a Cython version of the Kalman Filter.