Section: Software

Cuvclust package

Participant : Guillemette Marot.

cuvclust is a R package dedicated to model-based curve clustering. Considered models include Functional Clustering Mixed Models (FCMM, ie functional clustering with the presence of functional random effects), but also traditional functional clustering model (FCM, without functional random effects), and functional mixed models (FMM, functional random effects without clustering). Estimation is done by maximum likelihood using the EM algorithm, and two criteria are proposed to select the number of clusters, based on integrated likelihoods.

Guillemette Marot was the main contributor of the beta version of the package during her post-doc. Due to several changes in conception and due to planning of extensions in the package by the other contributors of the package, she decided to become a regular contributor and left the maintenance to Franck Picard.