idld_cluster_boot: IDLD Clustering with Bootstrap

In lfernandezpiana/idld: Integrated Dual Local Depth

IDLD Clustering with Bootstrap

Description

It is partition-based clustering technique based on local depth and distance measurement applied to data. This functional selects the optimal alpha_quantile for the procedure.

Usage

idld_cluster_boot(Z, beta, m, K, B, type, verbose = FALSE)

Arguments

Z	data to apply depth. It should be an array of dimension (n,p,l) where l is the number of functional coordinates. Z[,,i] is a numeric matrix where each row represents a functional observation for i=1,...,l.
beta	locality parameter between 0 and 1
m	number of random projections
K	number of clusters
B	number of bootstrap samples
type	the data type to apply the idld, "multivariate", "functional" or "multi_functional".
verbose	if TRUE prints the algorithm progress.

Value

returns a list with the following components:

local_depth a numeric vector object that contains the depth for each point.

region vector of booleans indicating which data points is in the central region.

clusters numeric vector with the partition.

Examples

library(funHDDC)
library(abind)
data("triangle")
triangle_data = abind(triangle[,1:101], triangle[,102:202], along=3)
d = dim(triangle_data)
triang_cl = idld_cluster_boot(triangle_data, 0.2, 100, 3, 20, "multi_functional", TRUE)
par(mfrow=c(1,2))
plot(triangle_data[1,,1], type="n", ylim=c(0,8))
for (i in 1:d[1]) lines(triangle_data[i,,1], col=triang_cl$clusters[i])
plot(triangle_data[1,,2], type="n", ylim=c(0,8))
for (i in 1:d[1]) lines(triangle_data[i,,2], col=triang_cl$clusters[i])

lfernandezpiana/idld documentation built on Feb. 17, 2024, 11:42 p.m.