get_covariance: Covariance estimate for latent CTMC parameters

In cthmm: Latent Continuous Time Markov Chains

Description

Obtain the covariance matrix for parameter estimates from the latent CTMC model.

Usage

  get_covariance(par, the.data, num.subjects, num.states,
    num.obs.states, rates.setup, emission.setup = NULL,
    init.setup = NULL, DDO.setup = NULL, do.DDO = F,
    time.dep.emission = F)

Arguments

par	the parameter estimates (rates, emission, initial distribution, DDO (in that order)
the.data	list with the observed data, with one entry per individual
num.subjects	number of individuals in the study
num.states	number of latent states in the CTMC
num.obs.states	number of observed states
rates.setup	list with rate setup information
emission.setup	list with emission distribution setup information
init.setup	list with initial distribution setup information
DDO.setup	setup object for disease driven observation model
do.DDO	indicator (T/F) if there are disease driven observation times in the model
time.dep.emission	(not supported) Indicator if emission distribution has time dependent covariates.

Details

Run after the EM algorithm that provides maximum likelihood estimates. Evaluates the hessian of the observed data likelihood, using numeric differentiation,

Value

covariance	Estimated covariance of parameter estiamtes
information	Observed Fisher information of parameter estimates

.

Author(s)

Jane Lange

Examples

## Not run: 
library(cthmm)
data(DDO_data)
#run the EM on the example data
covariance_DDO=get_covariance(par=DDO_EM$param,
the.data=DDO_data,
num.subjects=500,
num.states=4,
num.obs.states=3,
rates.setup=rates.setup,
emission.setup=emission.setup,
init.setup=init.setup,
DDO.setup=DDO.setup,
do.DDO=T)
## End(Not run)

cthmm documentation built on May 2, 2019, 5:46 p.m.