Naive implementation of the Kalman filter, smoother and disturbance smoother for state space models.
o The examples and code scripts have been updated now that the class and methods defined in package "stsm.class" have been merged in package "stsm".
o function "KalmanFilter": the function has been adapted to the new definition "KFAS::KFS" and it is again available; "sspir::kfilter" is not included because the archived package "sspir" is not available for "R Under devel" and may give some problems when submitting to CRAN and passing R CMD check. The call to "sspir" can nevertheless be found in 'KF-interfaces.R' prepended by comment marks.
o Added argument "xreg" to functions "KF.deriv" and "KF.deriv.C". The analytical derivatives with respect to the coefficients of regressors specified in the observation equation canbe evaluated. This is useful for example in package "stsm", where the analytical derivatives of the likelihood function can be obtained now for a model with external regressors.
o Added the method "predict" for objects of class "stsmSS" Those objects are returned by "stsm.class::char2numeric". The method is equivalent to "stats::predict.StructTS" but the forecasts of the components are also returned. In addition, it is straightforward to use for a model fitted by any of the maximum likelihood procedures available in package "stsm".
o Fixed the following warning returned by the compiler at some points in the code: "warning: ISO C++ forbids variable length array ‘df’ [-Wvla]"
o Removed suggested package "sspir" since it is no longer available-
o Removed suggested package "KFAS" since the examples using this package relied on a previous version of "KFAS".
o First version submitted to CRAN.