Section: New Results

Data Units Selection in Statistics

Participant : Christophe Biernacki.

Usually, the data unit definition is fixed by the practitioner but it can happen that he/her hesitates between several data unit options. In this context, it is highlighted that it is possible to embed data unit selection into a classical model selection principle. The problem is introduced in a regression context before to focus on the model-based clustering and co-clustering context, for data of different kinds (continuous, count, categorical). This work is now in revision for an international journal [36]. It has led also to three invitations as a plenary session speaker to international or national conferences (the US Classification Society Conference [27], the French Classification Society Conference [28], the StatLearn conference [20]).

It is a joint work with Alexandre Lourme from University of Bordeaux.