RcppHMM (version 1.2.2)

initGHMM: Random Initialization for a Hidden Markov Model with emissions modeled as continuous variables

Description

Function used to generate a hidden Markov model with continuous variables and random parameters. This method allows using the univariate version of a Gaussian Mixture Model when setting m = 1. The code for the methods with categorical values or discrete data can be viewed in "initHMM" and "initPHMM", respectively.

Usage

initGHMM(n,m)

Arguments

n

the number of hidden states to use.

m

the number of variables generated by the hidden states (Dimensionality of the bbserved vector).

Value

A "list" that contains the required values to specify the model.

Model

it specifies that the observed values are to be modeled as a Gaussian mixture model.

StateNames

the set of hidden state names.

A

the transition probabilities matrix.

Mu

a matrix of means of the observed variables (rows) in each states (columns).

Sigma

a 3D matrix that has the covariance matrix of each state. The number of slices is equal to the maximum number of hidden states.

Pi

the initial probability vector.

References

Cited references are listed on the RcppHMM manual page.

Examples

# NOT RUN {
n <- 3
m <- 5
model <- initGHMM(n, m)
print(model)
# }