hmmvbClust: Cluster data with HMM-VB

In HDclust: Clustering High Dimensional Data with Hidden Markov Model on Variable Blocks

Description

This function clusters dataset with HMM-VB. First, for each data point it finds an optimal state sequence using Viterbi algorithm. Next, it uses Modal Baum-Welch algorithm (MBW) to find the modes of distinct Viterbi state sequences. Data points associated the same modes form clusters. If different data sets are clustered using the same HMM-VB, clustering results of one data set can be supplied as a reference during clustering of another data set to produce aligned clusters.

Usage

hmmvbClust(data, model = NULL, control = clustControl(),
  rfsClust = NULL, nthread = 1, bicObj = NULL)

Arguments

data	A numeric vector, matrix, or data frame of observations. Categorical values are not allowed. If a matrix or data frame, rows correspond to observations and columns correspond to variables.
model	An object of class 'HMMVB' that contains trained HMM-VB obtained by the call to function hmmvbTrain.
control	A list of control parameters for clustering. The defaults are set by the call clustControl().
rfsClust	A list of parameters for the reference cluster that can be used for alignment. See HMMVBclust for details.
nthread	An integer specifying the number of threads used in clustering.
bicObj	An object of class 'HMMVBBIC' which stores results of model selection. If provided, argument model is ignored.

Value

An object of class 'HMMVBclust'.

See Also

HMMVB-class, HMMVBclust-class, hmmvbTrain

Examples

# cluster using trained HMM-VB
Vb <- vb(1, dim=4, numst=2)
set.seed(12345)
hmmvb <- hmmvbTrain(iris[,1:4], VbStructure=Vb)
clust <- hmmvbClust(iris[,1:4], model=hmmvb)
show(clust)
pairs(iris[,1:4], col=getClsid(clust))


# cluster using HMMVBBIC object obtained in model selection
Vb <- vb(1, dim=4, numst=1)
set.seed(12345)
modelBIC <- hmmvbBIC(iris[,1:4], VbStructure=Vb)
clust <- hmmvbClust(iris[,1:4], bicObj=modelBIC)
show(clust)
pairs(iris[,1:4], col=getClsid(clust))

HDclust documentation built on May 2, 2019, 9:20 a.m.