The importance of the scales in heterogeneous robust clustering

Authors:
L. A. García-Escudero;A. Gordaliza
Affiliations:
Dpto. de Estadística e I.O., Universidad de Valladolid, 47005 Valladolid, Spain;Dpto. de Estadística e I.O., Universidad de Valladolid, 47005 Valladolid, Spain
Venue:
Computational Statistics & Data Analysis
Year:
2007

Citing 4
Cited 2

Improved feasible solution algorithms for high breakdown estimation

Computational Statistics & Data Analysis
A fast algorithm for the minimum covariance determinant estimator

Technometrics
Influence function and efficiency of the minimum covariance determinant scatter matrix estimator

Journal of Multivariate Analysis
Validating visual clusters in large datasets: fixed point clusters of spectral features

Computational Statistics & Data Analysis

RelaxMCD: Smooth optimisation for the Minimum Covariance Determinant estimator

Computational Statistics & Data Analysis
The influence function of the TCLUST robust clustering procedure

Advances in Data Analysis and Classification

Quantified Score

Hi-index	0.03

Visualization

Abstract

The estimation of the scales plays an important role in the derivation of robust cluster techniques. This holds specially for some recently proposed methods including the so-called ''concentration'' steps in their implementation. A new robust clustering approach is introduced, the SSC method, intended to deal with scales, sizes and contamination. The method starts from a high trimming level which surely serves to remove all the outlying observations. Later, an iterative process is carried out where special attention is paid to the proper estimation of the groups' scales. The estimation of the contamination level is also considered.