HDclust (version 1.0.4)

hmmvbBIC: BIC for HMM-VB

Description

This function finds an optimal number of mixture components (states) for HMM-VB using the Bayesian Information Criterion (BIC). The variable block structure is provided as input and then BIC is estimated for HMM-VB with different configurations of states for the variable blocks.

Usage

hmmvbBIC(
  data,
  VbStructure,
  configList = NULL,
  numst = 1:10,
  trControl = trainControl(),
  nthread = 1
)

Value

A named list with estimated BIC values and the number of states or state configurations for which BIC was calculated.

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.

VbStructure

An object of class 'VB'. Variable block structure stored in VbStructure is used to train HMM-VB model. numst parameter of the variable block structure is ignored.

configList

A list of integer vectors specifying number of states in each variable block for which BIC is to be calculated.

numst

An integer vector specifying the numbers of mixture components (states) in each variable block for which BIC is to be calculated. Number of states is the same for all variable blocks. The argument is ignored if configList argument is provided.

trControl

A list of control parameters for HMM-VB training algorithm. The defaults are set by the call hmmvbTrainControl().

nthread

An integer specifying the number of threads used in searching and training routines.

See Also

VB, vb, trainControl

Examples

# \donttest{
# Default search for the optimal number of states for HMM-VB model 
data("sim3")
Vb <- vb(2, dim=40, bdim=c(10,30), numst=c(1,1), varorder=list(c(1:10),c(11:40)))
set.seed(12345)
hmmvbBIC(sim3[1:40], VbStructure=Vb)

# Search for the optimal number of states for HMM-VB model using 
# provided values for the number of states 
data("sim3")
Vb <- vb(2, dim=40, bdim=c(10,30), numst=c(1,1), varorder=list(c(1:10),c(11:40)))
set.seed(12345)
hmmvbBIC(sim3[1:40], VbStructure=Vb, numst=c(2L, 4L, 6L))

# Search for the optimal number of states for HMM-VB model using 
# provided configurations of the number of states 
data("sim3")
Vb <- vb(2, dim=40, bdim=c(10,30), numst=c(1,1), varorder=list(c(1:10),c(11:40)))
set.seed(12345)
configs = list(c(1,2), c(3,5), c(6,7))
hmmvbBIC(sim3[1:40], VbStructure=Vb, configList=configs)# }