4.8.6.4.2. statsmodels.tsa.kalmanf.kalmanfilter.StateSpaceModel¶

class statsmodels.tsa.kalmanf.kalmanfilter.StateSpaceModel(endog, exog=None, **kwargs)[source]¶

Generic StateSpaceModel class. Meant to be a base class.

This class lays out the methods that are to be defined by any child class.

Parameters:	endog : array-like An nobs x p array of observations exog : array-like, optional An nobs x k array of exogenous variables. **kwargs Anything provided to the constructor will be attached as an attribute.

Notes

The state space model is assumed to be of the form

y[t] = Z[t].dot(alpha[t]) + epsilon[t] alpha[t+1] = T[t].dot(alpha[t]) + R[t].dot(eta[t])

where

epsilon[t] ~ N(0, H[t]) eta[t] ~ N(0, Q[t]) alpha[0] ~ N(a[0], P[0])

Where y is the p x 1 observations vector, and alpha is the m x 1 state vector.

References

Durbin, J. and S.J. Koopman. 2001. `Time Series Analysis by State Space

Methods.` Oxford.

__init__(endog, exog=None, **kwargs)[source]¶