ziphsmm (version 2.0.6)

dist_learn3: Distributed learning for a longitudinal continuous-time zero-inflated Poisson hidden Markov model, where zero-inflation only happens in State 1 with covariates in the state-dependent parameters and transition rates.

Description

Distributed learning for a longitudinal continuous-time zero-inflated Poisson hidden Markov model, where zero-inflation only happens in State 1 with covariates in the state-dependent parameters and transition rates.

Usage

dist_learn3(ylist, xlist, timelist, M, initparm, yceil = NULL, rho = 1,
  priorclust = NULL, tpmclust = NULL, tpmslopeclust = NULL,
  emitclust = NULL, zeroclust = NULL, slopeclust = NULL, group,
  maxit = 100, tol = 1e-04, ncores = 1, seed = 0,
  method = "Nelder-Mead", print = TRUE, libpath = NULL, ...)

Arguments

ylist

list of observed time series values for each subject

xlist

list of design matrices for each subject.

timelist

list of time indices

M

number of latent states

initparm

matrix of initial working parameters for prior, transition, zero proportion, and emission parameters.

yceil

a scalar defining the ceiling of y, above which the values will be truncated. Default to NULL.

rho

tuning parameter in the distributed learning algorithm. Default to 1.

priorclust

a vector to specify the grouping for state prior. Default to NULL, which means no grouping.

tpmclust

a vector to specify the grouping for the intercepts in state transition rates. Default to NULL, which means no grouping.

tpmslopeclust

a vector to specify the grouping for the slopes in state transition rates. Default to NULL, which means no grouping.

emitclust

a vector to specify the grouping for the intercepts in Poisson regressions. Default to NULL, which means no grouping.

zeroclust

a vector to specify the grouping for the intercepts in ZIP regression. Default to NULL, which means no grouping.

slopeclust

a vector to specify the grouping for the slopes in Poisson and ZIP regressions. Default to NULL, which means no grouping.

group

a list containing group information.

maxit

maximum number iteration. Default to 100.

tol

tolerance in the terms of the relative change in the norm of the common coefficients. Default to 1e-4.

ncores

number of cores to be used for parallel programming. Default to 1.

seed

a seed for the random initialization of the algorithm

method

method for the distributed optimization in the ADMM framework.

print

whether to print each iteration. Default to TRUE.

libpath

path for the ziphsmm library if not the default set up. Default to NULL.

...

Further arguments passed on to the optimization methods

Value

the maximum likelihood estimates of the zero-inflated hidden Markov model

References

Boyd, S., Parikh, N., Chu, E., Peleato, B. and Eckstein, J., 2011. Distributed optimization and statistical learning via the alternating direction method of multipliers. Foundations and Trends in Machine Learning, 3(1), pp.1-122.

Examples

# NOT RUN {
set.seed(12933)
nsubj <- 10
ns <- 2000
ylist <- vector(mode="list",length=nsubj)
xlist <- vector(mode="list",length=nsubj)
timelist <- vector(mode="list",length=nsubj)

priorparm <- 0
tpmparm <- c(-2,0.1,-2,-0.2)
zeroindex <- c(1,0)
zeroparm <- c(0,0.5)
emitparm <- c(2,0.2,3,0.3)
workparm <- NULL

for(n in 1:nsubj){
 
 xlist[[n]] <- matrix(rep(c(0,1),rep(1000,2)),nrow=2000,ncol=1)
 
 timeindex <- rep(1,2000)
 for(i in 2:2000) timeindex[i] <- timeindex[i-1] + sample(1:4,1)
 timelist[[n]] <- timeindex
   
   workparm <- rbind(workparm,c(priorparm,tpmparm,zeroparm,emitparm))
 
 result <- hmmsim3.cont(workparm,2,2000,zeroindex,x=xlist[[n]],timeindex=timeindex)
 ylist[[n]] <- result$series
}


####
M <- 2
priorclust <- c(rep(1,5),rep(2,5))
tpmclust <- c(rep(1,5),rep(2,5))
tpmslopeclust <- c(rep(1,5),rep(2,5))
zeroclust <- NULL
emitclust <- NULL
slopeclust <- rep(1,10)

group <- vector(mode="list",length=2)
group[[1]] <- 1:5; group[[2]] <- 6:10
###
time <- proc.time()
result <- dist_learn3(ylist, xlist, timelist, 2,workparm, 
                     NULL, rho=1, priorclust,tpmclust,tpmslopeclust,
                     emitclust,zeroclust,slopeclust,group,ncores=1,
                     maxit=20, tol=1e-4, method="CG",print=TRUE)
proc.time() - time
# }