Oemparams: Emission Parameters

In NHMM: Bayesian Non-Homogeneous Markov and Mixture Models for Multiple Time Series

Description

Oemparams calculates emission parameters 0.025, 0.05, mean, 0.50 (median), 0.95, 0.975 quantiles from the iterations for each parameter. Each of the J sequences and K states each has 2 parameters (Gamma and Normal) or 1 parameter (Poisson) per mixture component (nmix+delta)

Usage

Oemparams(nhmmobj, plots = FALSE, outfile = NULL)

Arguments

nhmmobj	an object created from the NHMM or HMM function
plots	TRUE/FALSE- default is FALSE because the plot window can grow quite large depending on the number of parameters. [J by K*nmix] panes if outboo=TRUE and [K*nmix by J] panes if outboo=FALSE, where outboo is a parameter from NHMM which determines if output was written to a file (TRUE) or to a variable (FALSE). If outfile is used then there will be K*J .png files containing dimension 2 parameters by nmix panes of trace plots. Exception: NHMM_MVN, plots is always FALSE.
outfile	a directory to put the .png plot

Value

params [2 by nmix by K by J] by 6. There are six values returned: 0.025, 0.05, mean, 0.50 (median), 0.95, 0.975 quantiles from the iterations. (0.025, 0.975) are used to construct 95 likewise (0.05, 0.95) can be used to construct 90 but the one-parameter distributions like the Exponential will only have meaningful data in the first row. Exception for NHMM_MVN then the parameters returned is the mean covariance matrix over all iterations [J by J by K]

output: plot window can grow quite large depending on the number of parameters. [J by K*nmix] panes if outboo=TRUE and [K*nmix by J] panes if outboo=FALSE, where outboo is a parameter from NHMM which determines if output was written to a file (TRUE) or to a variable (FALSE). Exception: NHMM_MVN object does not plot the covariance.

Examples

#thetas=Oemparams(my.nhmm, FALSE); 
#thetas[,,,,3]  #mean values

NHMM documentation built on July 1, 2020, 7:28 p.m.