find_clusters {cvxclustr}R Documentation

Find clusters

Description

find_clusters uses breadth-first search to identify the connected components of the corresponding adjacency graph of the centroid differences vectors.

Usage

find_clusters(A)

Arguments

A

adjacency matrix

Examples

## Clusterpaths for Mammal Dentition
data(mammals)
X <- as.matrix(mammals[,-1])
X <- t(scale(X,center=TRUE,scale=FALSE))
n <- ncol(X)

## Pick some weights and a sequence of regularization parameters.
k <- 5
phi <- 0.5
w <- kernel_weights(X,phi)
w <- knn_weights(w,k,n)
gamma <- seq(0.0,43, length.out=100)

## Perform clustering
nu <- AMA_step_size(w,n)
sol <- cvxclust_path_ama(X,w,gamma,nu=nu)

## Construct adjacency matrix
A <- create_adjacency(sol$V[[10]],w,n)
find_clusters(A)

[Package cvxclustr version 1.1.1 Index]