Clustering variables: a survey and some new developments - Cnam - Conservatoire national des arts et métiers Accéder directement au contenu
Communication Dans Un Congrès Année : 2024

Clustering variables: a survey and some new developments

Résumé

Alongside the profusion of clustering methods for observations, there are few specific methods for clustering variables. This presentation will provide an overview of the methods that can be used: partitioning, hierarchical methods and mixture models, with a particular focus on the case of sets combining quantitative and qualitative variables. The use of p-values to eliminate the effects due to different numbers of categories is often recommended, but when the number of units is very large, all p-values are close to zero (paradox of large samples). We propose to replace them by z-scores. These different approaches will be compared on real data sets.
Fichier non déposé

Dates et versions

hal-04601242 , version 1 (04-06-2024)

Identifiants

  • HAL Id : hal-04601242 , version 1

Citer

Ndèye Niang, Mory Ouattara, Gilbert Saporta. Clustering variables: a survey and some new developments. Sensometrics 2024, The Sensometric Society, Jun 2024, Paris, France. ⟨hal-04601242⟩
14 Consultations
0 Téléchargements

Partager

Gmail Mastodon Facebook X LinkedIn More