move.HSMM.viterbi: Assign states using the Viterbi algorithm

In benaug/move.HMM: Fit HMM and HSMM animal movement models

Description

This function, modified from Zucchini and MacDonald (2009), assigns states to observations using the Viterbi algorithm. It takes as input a move.HSMM object and an optional vector containing the starting state probabilities.

Usage

  move.HSMM.viterbi(move.HSMM, delta = NULL)

Arguments

move.HSMM	A move.HSMM object containing a fitted HSMM model.
delta	An optional vector of starting state probabilities. If no vector is supplied, the stationary distribution is used.

Value

A vector of state assignments.

Examples

## Not run: 
#2 states, 2 dist-lognorm, wrapped normal
lmean=c(-3,-1) #meanlog parameters
sd=c(1,1) #sdlog parameters
rho<-c(0.2,0.3) # wrapped normal concentration parameters
mu<-c(pi,0) # wrapped normal mean parameters
gamma0=matrix(c(0.6,0.4,0.2,0.8),byrow=T,nrow=2)

dists=c("lognormal","wrpnorm")
nstates=2
turn=c(1,2)
params=vector("list",3)
params[[1]]=gamma0
params[[2]]=cbind(lmean,sd)
params[[3]]=cbind(mu,rho)
obs=move.HSMM.simulate(dists,params,1000)$obs
turn=c(1,2)
move.HSMM=move.HSMM.mle(obs,dists,params,stepm=35,iterlim=100,turn=turn)
#get Viterbi state assignments
move.HSMM.viterbi(move.HSMM)

## End(Not run)

benaug/move.HMM documentation built on Jan. 23, 2022, 4:29 a.m.