Skip to contents

Hierarchical Clustering

Source: R/mv_stats.R
mv_hclust.Rd

Standardizes selected numeric columns (optional), builds a distance matrix, runs hierarchical clustering (stats::hclust), and cuts the tree into k clusters.

Usage

mv_hclust(
  data,
  cols,
  k = 3,
  dist_method = "euclidean",
  method = "ward.D2",
  scale = TRUE
)

Arguments

data

A data frame.

cols

<tidy-select> Numeric columns to cluster on.

k

Number of clusters to extract by cutting the tree.

dist_method

Distance method passed to stats::dist, e.g. "euclidean" (default), "maximum", "manhattan", "canberra", "binary", or "minkowski".

method

Agglomeration method passed to stats::hclust, e.g. "ward.D2" (default), "ward.D", "single", "complete", "average", "mcquitty", "median", or "centroid".

scale

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

Value

An object of class mv_hclust: a list with components cluster (tibble with .row and cluster), heights (tibble with step, join_1, join_2, height; join_1/join_2 use hclust merge encoding — negative values index observation leaves, positive values index earlier merged clusters), hclust (the raw hclust object, useful for plot()), data_mat (the standardized analysis matrix), and meta.

Examples

r = mv_hclust(mtcars, cyl:carb, k = 3)
table(r$cluster$cluster)
#> 
#>  1  2  3 
#>  5 15 12 
plot_mv_dendrogram(r)