Automatically find cluster assignment given f and delta.

Share:

Description

This is the subroutine that automatically finds cluster assignments from given f and delta by testing various parameter settings and find the one that maximizes the silhouette.

Usage

1
2
FindClustersAuto(distm, f, delta, ac = 1, nclust = 2:10, f.cut = c(0.1,
  0.2, 0.3))

Arguments

distm

the distance matrix

f

vector of local distance f(x). See the help of adpclust() for details.

delta

vector of minimal distances to higher ground delta(x). See the help of adpclust() for details.

ac

type of auto selection. The valid options are 1 and 2. See the help of adpclust() for details.

nclust

number of clusters to test. Either a single integer or a vector of integers.

f.cut

number between (0, 1) or numeric vector of numbers between (0, 1). Data points whose f values are larger than f.cut with large delta values are selected as centers. The default is c(0.1, 0.2, 0.3). See the help of FindCentersAutoV() for more details.

Value

list of four elements:

  • clusters Cluster assignments. A vector of the same length as the number of observations.

  • centers: Indices of the clustering centers.

  • silhouette: Silhouette score from the final clustering result.

  • nclust: Number of clusters.

Author(s)

Ethan Xu