kmeans_init: Choose Initial K-Means Values

Description Usage Arguments Value Examples

Description

Choose Initial K-Means Values

Usage

1
kmeans_init(data = NULL, K = NULL, method = "kmeanspp", seed = NULL)

Arguments

data

the data object (data frame or matrix) that k-means clustering will be applied to.

K

the number of initial values to be chosen. Should correspond to the number of clusters to be chosen.

method

"kmeanspp" or "rp". The initialisation method specified as a string. "kmeanspp" refers to kmeans++ method and "rp" refers to random points method. More information on kmeans initialization methods can be found here.

seed

the seed to be set if "rp" is specified as method. If NULL, no seed will be set.

Value

A matrix with coordinates for initialization values, where each row is an initialization value and the columns correspond with the columns of the input data object.

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
# create input data object with two distinct clusters
data <- data.frame(x = runif(100, min = 0, max = 10) + rep(c(0, 10), 50), y = rnorm(100, 5, 1) + rep(c(0, 10), 50))

# kmeans++ algorithm by default
kmeans_init(data = data, K = 2)

# random points initialization method
kmeans_init(data = data, K = 2, method = "rp")

# random points initialization method with seed set
kmeans_init(data = data, K = 2, method = "rp", seed = 1234)

UBC-MDS/kmeans_R documentation built on May 22, 2019, 2:26 p.m.