E.step.emission.init.all: E step for emission and initial distributions

In cthmm: Latent Continuous Time Markov Chains

Description

Get the expected emission counts E[(O_ij)|o] and initial distribution counts E[(Z_i)|o] given observed data for all individuals. This implementation does not use the recursive filtering method, but rather uses forward and backward probabilities.

Usage

  E.step.emission.init.all(likelihood.forward.backward.list,
    the.data, num.states, num.obs.states, num.subjects)

Arguments

likelihood.forward.backward.list	list with LL, forward, and backward probs for each inividual
the.data	list with suject level obs.data
num.states	number of hidden states in the model
num.obs.state	number of observed states in the model
num.subjects	number of individuals

Value

a list with two elements:

emission.array	(dim (num.states, num.obs.states, num.subjects))
init.array	(dim (num.states,num.subjects))

Author(s)

Jane Lange

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