ssm {sspir} | R Documentation |
Use a glm-style formula and family arguments to setup a state space model.
ssm(formula, family = gaussian, data = list(), subset = NULL, fit = TRUE, phi = NULL, m0 = NULL, C0 = NULL, Fmat = NULL, Gmat = NULL, Vmat = NULL, Wmat = NULL) ## S3 method for class 'ssm': C0(ssm) ## S3 method for class 'ssm': m0(ssm) ## S3 method for class 'ssm': Fmat(ssm) ## S3 method for class 'ssm': Gmat(ssm) ## S3 method for class 'ssm': Vmat(ssm) ## S3 method for class 'ssm': Wmat(ssm) ## S3 method for class 'ssm': phi(ssm) ## S3 method for class 'ssm': C0(ssm) <- value ## S3 method for class 'ssm': m0(ssm) <- value ## S3 method for class 'ssm': Fmat(ssm) <- value ## S3 method for class 'ssm': Gmat(ssm) <- value ## S3 method for class 'ssm': Vmat(ssm) <- value ## S3 method for class 'ssm': Wmat(ssm) <- value ## S3 method for class 'ssm': phi(ssm) <- value getFit(ssm)
formula |
a formula with univariate response on the lefthand
side. The righthand side is a sum of terms and the special functions
sumseason , polytime ,
polytrig , and season can be used. Terms
can be marked by the tvar -function to create a term with
time-varying coefficients. A special case is tvar(1) meaning a
random walk. |
family |
a description of the error distribution and link function to
be used in the model. This can be a character string naming a
family function, a family function or the result of a call to
a family function. (See family for details of
family functions.) |
data |
an optional data frame containing the variables in the model.
If not found in data , the variables are taken from
environment(formula) , typically the environment from which
ssm is called. |
subset |
an optional vector specifying a subset of observations to be used in the fitting process. |
ssm |
an object of class ssm . |
fit |
a logical. If TRUE , the model is fit using the
kfs method. Otherwise, the model is returned as
is. |
phi |
a vector of initial values of hyperparamters. Note
that these have to be in the right order. Best advice is to
leave this option to be NULL and then inspect the
returned result using phi(ssm) . |
m0 |
a vector with the initial state vector. |
C0 |
a matrix with the variance matrix of the initial state. |
Fmat |
a function giving the regression matrix at a given timepoint. |
Gmat |
a function giving the evolution matrix at a given timepoint. |
Wmat |
a function giving the evolution variance matrix at a given timepoint. |
Vmat |
a function giving the observation variance matrix at a given timepoint. |
value |
an object to be assigned to the element of the state space model. |
An object of class ssm
with the following components
ss |
an object of class SS describing the state
space model. In addition, the ss object contains the
components family and ntotal (for binomial case). |
Claus Dethlefsen and Søren Lundbye-Christensen.
data(vandrivers) vandrivers$y <- ts(vandrivers$y,start=1969,frequency=12) vd.time <- time(vandrivers$y) vd <- ssm( y ~ tvar(1) + seatbelt + sumseason(vd.time,12), family=poisson(link="log"), data=vandrivers, phi = c(1,0.0004), C0=diag(13)*100, fit=FALSE ) phi(vd)["(Intercept)"] <- exp(- 2*3.703307 ) C0(vd) <- diag(13)*1000 vd.res <- kfs(vd) plot( vd.res$m[,1:3] ) attach(vandrivers) plot(y,ylim=c(0,20)) lines(exp(vd.res$m[,1]+vd.res$m[,2]*seatbelt),lwd=2 ) detach(vandrivers)