K-means Clustering

Standardizes selected numeric columns (optional) and runs stats::kmeans, returning tidy cluster assignments, centers, and within-cluster sums of squares.

Usage

mv_kmeans(data, cols, centers, nstart = 25, scale = TRUE, iter.max = 100)

Arguments

data

A data frame.

cols

<tidy-select> Numeric columns to cluster on.

centers

Either the number of clusters (k, a single integer) or a numeric matrix of initial cluster centers with the same number of columns as cols, passed to stats::kmeans. When a named matrix of centers is supplied, its row names are kept as cluster labels.

nstart

Number of random starts passed to stats::kmeans; ignored when centers is a matrix.

scale

Logical. If TRUE (default), columns are standardized first.

iter.max

Maximum iterations passed to stats::kmeans.

Value

An object of class mv_kmeans: a list with components cluster (tibble with .row and cluster), centers (tibble with cluster and one column per variable), within_ss (tibble with cluster, size, withinss, share), data_mat (the standardized analysis matrix), and meta.

Examples

r = mv_kmeans(iris, Sepal.Length:Petal.Width, centers = 3)
r$within_ss
#> # A tibble: 3 × 4
#>   cluster  size withinss share
#>   <fct>   <int>    <dbl> <dbl>
#> 1 1          47     47.5 0.342
#> 2 2          53     44.1 0.317
#> 3 3          50     47.4 0.341
plot_mv_silhouette(r)