clustering_metrics: Internal evaluation of clustering results

In vittoriofortino84/COPS: Clustering algorithms for Omics-driven Patient Stratification

Internal evaluation of clustering results

Description

Single embedding or dataset evaluation

Usage

clustering_metrics(
  x,
  dat = NULL,
  by = c("k", "m"),
  dissimilarity = NULL,
  cluster_size_table = TRUE,
  silhouette_min_cluster_size = 0,
  internal_metrics = NULL,
  ...
)

Arguments

x	A data.frame of clustering results.
dat	A data.frame with data columns identified with "dim". Not required if dissimilarity is defined.
by	vector of variable names to split by
dissimilarity	a dist object to use in silhouette calculation, defaults to euclidean distance matrix if left NULL
cluster_size_table	return cluster sizes if TRUE.
silhouette_min_cluster_size	proportional cluster size threshold for merging into nearest neighbours' cluster for silhouette computation.
internal_metrics	Internal metric names passed to intCriteria. This will slow the pipeline down considerably.
...	extra arguments are passed to clustering_dissimilarity_from_data

Value

Returns a list containing metrics and cluster sizes

vittoriofortino84/COPS documentation built on Jan. 28, 2025, 3:16 p.m.