B1_DEFAULT: ktaucenters_run
In anevolbap/ktaucenterscpp: Robust Clustering Procedures

Description Usage Arguments Format Value Note References See Also Examples

Robust Clustering algorithm based on centers, a robust and efficient version of K-Means.

1	B1_DEFAULT

`data`	A matrix of size n x p.
`K`	The number of clusters.
`centers`	matrix of size K x p containing the K initial centers, one at each matrix-row.
`tolerance`	tolerance parameter used for the algorithm stopping rule
`max_iter`	a maximum number of iterations used for the algorithm stopping rule

An object of class numeric of length 1.

A list including the estimated K centers and labels for the observations

centers: matrix of size K x p, with the estimated K centers.
cluster: array of size n x 1 integers labels between 1 and K.
tauPath: sequence of tau scale values at each iterations.
Wni: numeric array of size n x 1 indicating the weights associated to each observation.
empty_cluster_flag: a boolean value. True means that in some iteration there were clusters totally empty
niter: number of iterations until convergence is achived or maximum number of iteration is reached
didistance of each observation to its assigned cluster-center

Some times, if the initial centers are wrong, the algorithm converges to a non-optimal (local) solution. To avoid that, the algorithm must be run several times. This task is carried out by ktaucenters

Gonzalez, J. D., Yohai, V. J., & Zamar, R. H. (2019). Robust Clustering Using Tau-Scales. arXiv preprint arXiv:1906.08198.

ktaucenters

# Generate Synthetic data (three cluster well separated)
Z=rnorm(600);
mues=rep(c(0,10,20),200)
data= matrix(Z+mues,ncol=2)

# Applying the algorithm
sal = ktaucenters_aux(
data, K=3, centers=data[sample(1:300,3), ],
tolerance=1e-3, max_iter=100)

#plot the results
plot(data,type='n')
points(data[sal$cluster==1,],col=1);
points(data[sal$cluster==2,],col=2);
points(data[sal$cluster==3,],col=3);