smooth.discrete: Smoothing a discrete time series.

In jaredo/mhsmm: Inference for Hidden Markov and Semi-Markov Models

Description

The smooth.discrete() function provides a simple smoothing of a time series of discrete values measured at equidistant times. Under the hood of smooth.discrete() is a hidden Markov model.

Usage

smooth.discrete(y, init = NULL, trans = NULL, parms.emission = 0.5, 
                method = "viterbi", details = 0, ...)

Arguments

y	A numeric vector
init	Initial distribution (by default derived from data; see the vignette for details)
trans	Transition matrix (by default derived from data; see the vignette for details)
parms.emission	Matrix describing the conditional probabilities of the observed states given the latent states. (See the vignette for details).
method	Either "viterbi" or "smoothed". The viterbi method gives the jointly most likely sequence; the smoothed method gives the sequence of individually most likely states.
details	Controlling the amount of information printed.
...	Further arguments passed on to the "hmmfit" function.

Details

The parameters are estimated using the Baum-Welch algorithm (a special case of the EM-algorithm).

Value

A list with the following components:

s	The "smoothed" states
model	The underlying hmm (hidden Markov model) object
data	The data
initial	The initial parameters

Author(s)

Søren Højsgaard <sorenh at agrsci.dk>

See Also

hmmspec, hmmfit

Examples

## Please see the vignette

jaredo/mhsmm documentation built on Dec. 6, 2019, 11:07 a.m.