orthoKMeansTrain: Orthogonal k-Means training.

Description Usage Arguments Value Examples

Description

orthoKMeansTrain will cluster a given data set into the specified number of clusters. It can use either random initialization of the centroids or use KMeans++ for this. The K-Means training itself is accelerated by using techniques by Greg Hamerly. Orthoginality is implemented by using ideas from Cui et al 'Non-redudant multi-view clustering via orthogonalization'.

Usage

1
2
orthoKMeansTrain(x = NULL, k = NULL, rounds = 1, iter.max = 100,
  init.type = "KMeans++", verbose = FALSE)

Arguments

x

data to cluster

k

number of centroids

rounds

number of rounds/views for orthogonal kmeans

iter.max

number of maximal iterations for each clustering

init.type

string with method to initialize centroids

verbose

show verbose messages?

Value

an S3 object containing the cluster labels for the training set as well as all necessary information for prediction.

Examples

1
2
obj = yakmoR::orthoKMeansTrain (x = as.matrix(iris[seq(1,150,2),1:4]),
			k = 3, rounds = 3, verbose = TRUE)

yakmoR documentation built on May 1, 2019, 8:41 p.m.