hsmmviterbi: Viterbi algorithm to decode the latent states for hidden...

In ziphsmm: Zero-Inflated Poisson Hidden (Semi-)Markov Models

Description

Viterbi algorithm to decode the latent states for hidden semi-Markov models

Usage

hsmmviterbi(y, ntimes = NULL, M, trunc, prior, dt_dist = "nonparametric",
  dt_parm, tpm_parm, emit_parm, zero_init, plot = TRUE, xlim = NULL,
  ylim = NULL, ...)

Arguments

y	the observed series to be decoded
ntimes	vector specifying the lengths of individual, i.e. independent, time series. If not specified, the responses are assumed to form a single time series, i.e. ntimes=length(y)
M	number of latent states
trunc	a vector specifying truncation at the maximum number of dwelling time in each state.
prior	a vector of prior probability values
dt_dist	dwell time distribution, can only be "log", "shiftedpoisson", or "nonparametric". Default to "nonparametric".
dt_parm	a vector of parameter values in each dwell time distribution, which should be a vector of p's for dt_dist == "log", or a vector of theta's for dt_dist=="shiftpoisson", or a matrix whose a matrix whose i,j th element is the probability of staying in state i for duration j for dt_dist == "nonparametric".
tpm_parm	transition probability matrix
emit_parm	a vector containing means for each poisson distribution
zero_init	a vector containing structural zero proportions in each state
plot	whether a plot should be returned
xlim	vector specifying the minimum and maximum on the x-axis in the plot. Default to NULL.
ylim	vector specifying the minimum and maximum on the y-axis in the plot. Default to NULL.
...	further arguments to be passed to the plot() function

Value

the decoded series of latent states

References

Walter Zucchini, Iain L. MacDonald, Roland Langrock. Hidden Markov Models for Time Series: An Introduction Using R, Second Edition. Chapman & Hall/CRC

Examples

## Not run: 
#3 zero-inflated poissons
prior_init <- c(0.3,0.3,0.4)
dt_init <- c(10,8,6)
emit_init <- c(10,50,100)
zeroprop <- c(0.5,0.3,0.2)
trunc <- c(10,10,10)
omega <- matrix(c(0,0.3,0.7,0.4,0,0.6,0.5,0.5,0),3,3,byrow=TRUE)
result <- hsmmsim(n=1000,M=3,prior=prior_init,dt_dist="shiftpoisson",
         dt_parm=dt_init, tpm_parm=omega,emit_parm=emit_init,zeroprop=zeroprop)
y <- result$series
state <- result$state
fit <- hsmmfit(y=y,ntimes=NULL,M=3,trunc=trunc,prior_init=prior_init,dt_dist="shiftpoisson",
     dt_init=dt_init,tpm_init=omega,emit_init=emit_init,zero_init=zeroprop,
     method="Nelder-Mead",hessian=FALSE,control=list(maxit=500,trace=1))
decode <- hsmmviterbi(y=y,ntimes=NULL,M=3,trunc=trunc,prior=fit$prior,dt_dist="shiftpoisson",
     dt_parm=fit$dt_parm,tpm_parm=fit$tpm,emit_parm=fit$emit_parm,
     zero_init=fit$zeroprop,plot=TRUE,xlim=c(0,1000),ylim=c(0,200))
#check the missclassification rate
sum(decode!=state)/length(state)

## End(Not run)   
     

ziphsmm documentation built on May 2, 2019, 6:10 a.m.