mixtureProbs {momentuHMM}R Documentation

Mixture probabilities

Description

For a fitted model, this function computes the probability of each individual being in a particular mixture

Usage

mixtureProbs(m, getCI = FALSE, alpha = 0.95)

Arguments

m

momentuHMM or momentuHierHMM object

getCI

Logical indicating whether to calculate standard errors and logit-transformed confidence intervals for fitted momentuHMM or momentuHierHMM object. Default: FALSE.

alpha

Significance level of the confidence intervals (if getCI=TRUE). Default: 0.95 (i.e. 95% CIs).

Details

When getCI=TRUE, it can take a while for large data sets and/or a large number of mixtures because the model likelihood for each individual must be repeatedly evaluated in order to numerically approximate the SEs.

Value

The matrix of individual mixture probabilities, with element [i,j] the probability of individual i being in mixture j

References

Maruotti, A., and T. Ryden. 2009. A semiparametric approach to hidden Markov models under longitudinal observations. Statistics and Computing 19: 381-393.

Examples

## Not run: 
nObs <- 100
nbAnimals <- 20
dist <- list(step="gamma",angle="vm")
Par <- list(step=c(100,1000,50,100),angle=c(0,0,0.1,2))

# create sex covariate
cov <- data.frame(sex=factor(rep(c("F","M"),each=nObs*nbAnimals/2)))
formulaPi <- ~ sex + 0

# Females more likely in mixture 1, males more likely in mixture 2
beta <- list(beta=matrix(c(-1.5,-0.5,-1.5,-3),2,2),
             pi=matrix(c(-2,2),2,1,dimnames=list(c("sexF","sexM"),"mix2"))) 

data.mix<-simData(nbAnimals=nbAnimals,obsPerAnimal=nObs,nbStates=2,dist=dist,Par=Par,
                  beta=beta,formulaPi=formulaPi,mixtures=2,covs=cov) 

Par0 <- list(step=Par$step, angle=Par$angle[3:4])   
m.mix <- fitHMM(data.mix, nbStates=2, dist=dist, Par0 = Par0, 
                beta0=beta,formulaPi=formulaPi,mixtures=2)
                
mixProbs <- mixtureProbs(m.mix, getCI=TRUE)

## End(Not run)

[Package momentuHMM version 1.5.5 Index]