clustControl: Parameters for MBM clustering algorithm.

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

Description

This function creates a list with parameters for Modal Baum-Welch (MBW) clustering algorithm used as an argument for hmmvbClust.

Usage

clustControl(minSize = 1, modeTh = 0.01, useL1norm = FALSE,
  getlikelh = FALSE)

Arguments

minSize	Minimum cluster size. Clusters that contain the number of data points smaller than minSize are merged to the closest big cluster.
modeTh	Distance parameter that controls mode merging. Larger values promote merging of different clusters.
useL1norm	A logical value indicating whether or not L1 norm will be used to calculate the distance.
getlikelh	A logical value indicating whether or not to calculate the loglikelihood for every data point.

Value

The named list with parameters.

See Also

hmmvbTrain

Examples

# avoid clusters of size < 60
Vb <- vb(1, dim=4, numst=2)
set.seed(12345)
hmmvb <- hmmvbTrain(iris[,1:4], VbStructure=Vb)
clust <- hmmvbClust(iris[,1:4], model=hmmvb, control=clustControl(minSize=60))
show(clust)

Example output

Data size: 150 number of large clusters: 1 #points in large clusters: 100
Warning: percentage of points in large clusters is small: %66.6667 < 80 percent
------------------------------------------------------
Clustering with Hidden Markov Model on Variable Blocks
------------------------------------------------------

Number of clusters = 2 

Cluster sizes: 0 150 

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