Section: Software

Two advances for the MIXMOD software

Participants : Christophe Biernacki, Serge Iovleff, Remi Lebret, Parmeet Bhatia.

MIXMOD (MIXture MODelling) is an important software for the m$\Theta$dal team since it concerns its main topics: model-based supervised, unsupervised and semisupervised classification for various data situations. MIXMOD is now a well-distributed software with over 250 downloads/month are recorded for several years. MIXMOD is written in C++ (more than 10 000 lines) and distributed under GNU General Public License. Several other institutions participate in the MIXMOD development since several years: CNRS, Inria Saclay-Île de France, Université de Franche-Comté, Université Lille 1. The software already benefits from several APP deposits.

An interface between MIXMOD and R (Rmixmod) has been developed by Rémi Lebret and Serge Iovleff and is now available on the CRAN (http://cran.r-project.org/web/packages/Rmixmod/index.html). We expect now a wide impact of MIXMOD on the growing community familiar with R. A paper related to Rmixmod is submitted to an international journal [34] .

Until December 2012, Parmeet Bhatia, under scientific supervision of Christophe Biernacki, is developping possibility in MIXMOD to cluster simultaneously continuous and categorical data with the restrictive conditional independence assumption. It is an important first step towards the long term purpose of m$\Theta$dal to cluster heterogeneous (or mixed) data sets.