ClustOfVar : un package R pour la classification de variables

CHAVENT, Marie; KUENTZ, V.; LIQUET, B.; SARACCO, Jérôme

Métadonnées

Afficher la notice complète

Licence d’utilisation du document

CHAVENT, Marie
Institut de Mathématiques de Bordeaux [IMB]

KUENTZ, V.
Aménités et dynamiques des espaces ruraux [UR ADBX]

LIQUET, B.
Université de Bordeaux [UB]

Langue

Communication dans un congrès

Ce document a été publié dans

The R User Conference 2011, University of Warwick, 2011-08-16, Coventry. 2011p. 1

Résumé en anglais

Clustering of variables is as a way to arrange variables into homogeneous clusters i.e. groups of variables which are strongly related to each other and thus bring the same information. Clustering of variables can then be useful for dimension reduction and variable selection. Several specific methods have been developed for the clustering of numerical variables. However concerning qualitative variables or mixtures of quantitative and qualitative variables, much less methods have been proposed. The ClustOfVar package has then been developped specifically for that purpose. The homogeneity criterion of a cluster is the sum of correlation ratios (for qualitative variables) and squared correlations (for quantitative variables) to a synthetic variable, summarizing as good as possible the variables in the cluster. This synthetic variable is the first principal component obtained with the PCAMIX method. Two algorithms for the clustering of variables are proposed: iterative relocation algorithm, ascendant hierarchical clustering. We also propose a bootstrap approach in order to determine suitable numbers of clusters. The proposed methodologies are illustrated on real datasets.< Réduire

Métadonnées

Partager cette publication !

Licence d’utilisation du document