speccalt: speccalt

In speccalt: Alternative spectral clustering, with automatic estimation of k.

Description

Spectral clustering algorithm, with optional automatic estimation of the optimal number of clusters.

Usage

speccalt(kern, k=NA, maxk=20)

Arguments

kern	A kernel matrix object, as returned e.g. by local.rbfdot.
k	Parameterized number of clusters. If NA, the automatic estimation procedure (bartlett) is used.
maxk	If k is NA, an upper bound for the automatic estimation. Defaults to 20.

Value

A numeric vector of cluster labels. Clusters are indexed from 1 to k.

Author(s)

Pierrick Bruneau

See Also

local.rbfdot bartlett

Examples

kern <- local.rbfdot(synth2)
clust <- speccalt(kern) # with automatic estimation
clust2 <- speccalt(kern, 4) # without automation

Example output

speccalt documentation built on May 2, 2019, 8:53 a.m.