predict.hmm.discnp: Predicted values of a discrete non-parametric hidden Markov...

In hmm.discnp: Hidden Markov Models with Discrete Non-Parametric Observation Distributions

Predicted values of a discrete non-parametric hidden Markov model.

Description

Calculates predicted values given a specification of a discrete non-parametric hidden Markov model. The specification may be provided in the form of a hmm.discnp object as returned by hmm() or in the form of “components” of such a model: the data y, the transition probability matrix tpm, the emission probabilities Rho, etc. If the data are numeric then these predicted values are the conditional expectations of the observations, given the entire observation sequence (and the — possibly estimated — parameters of the model). If the data are categorical (whence “expectations” make no sense) the “predicted values” are taken to be the probabilities of each of the possible values of the observations, at each time point.

Usage

## S3 method for class 'hmm.discnp'
predict(object, y = NULL, tpm=NULL, Rho=NULL,
                             ispd=NULL, X=NULL,addIntercept=NULL,
                             warn=TRUE, drop=TRUE, ...)

Arguments

object	If not NULL, an object of class hmm.discnp as returned by hmm().
y	A data structure to which the fitted model object could have been fitted. If y is NULL, an attempt is made to extract y from model.
tpm,Rho,ispd,X,addIntercept,warn	See the help for sp().
drop	Logical scalar. See the help for fitted.hmm.discnp().
...	Not used.

Details

This function is essentially the same as fitted.hmm.discnp(). The main difference is that it allows the calculation of fitted/predicted values for a data object y possibly different from that to which the model was fitted. Note that if both the argument y and object[["y"]] are present, the “argument” value takes precedence. This function also allows the model to be specfied in terms of individual components rather than as a fitted model of class "hmm.discnp". These components, (tpm, Rho, ispd, X, addIntercept) if supplied, take precedence over the corresponding components of object. The opposite applies with sp(). The function fitted.hmm.discnp() makes use only of the components of object.

Value

See the help for fitted.hmm.discnp().

Author(s)

Rolf Turner [e-mail]

See Also

sp() link{fitted.hmm.discnp}()

Examples

P <- matrix(c(0.7,0.3,0.1,0.9),2,2,byrow=TRUE)
R <- matrix(c(0.5,0,0.1,0.1,0.3,
              0.1,0.1,0,0.3,0.5),5,2)
set.seed(42)
ll1  <- sample(250:350,20,TRUE)
y1   <- rhmm(ylengths=ll1,nsim=1,tpm=P,Rho=R,drop=TRUE)
fit  <- hmm(y1,K=2,verb=TRUE,keep.y=TRUE,itmax=10)
fv   <- fitted(fit)
set.seed(176)
ll2  <- sample(250:350,20,TRUE)
y2   <- rhmm(ylengths=ll2,nsim=1,tpm=P,Rho=R,drop=TRUE)
pv   <- predict(fit,y=y2)
yval <- letters[1:5]
set.seed(171)
y3   <- rhmm(ylengths=ll2,yval=yval,nsim=1,tpm=P,Rho=R,drop=TRUE)
fit3 <- hmm(y3,K=2,verb=TRUE,keep.y=TRUE,itmax=10)
pv3  <- predict(fit3) # Same as fitted(fit3).

hmm.discnp documentation built on Sept. 26, 2022, 5:05 p.m.