**Speaker:** Vincent Mousseau (Professeur à CentraleSupélec, Laboratoire de Génie Industriel)

**Title** Learning monotone preferences using a non-compensatory sorting model,

*son travail en collaboration avec Khaled Belahcene, Eda Ersek, Wassila Ouerdane, Marc Pirlot et Olivier Sobrie.*

**Abstract**

We consider the problem of learning a function assigning objects into ordered categories. The objects are described by a vector of attribute values and the assignment function is monotone w.r.t. the attribute values (monotone sorting problem). Our approach is based on a model used in Multi-Criteria Decision Analysis (MCDA), called Non Compensatory Sorting (NCS). This model assigns an object a to a class when a is better than the lower frontier of the class on a “sufficient” set of criteria, and this is not true w.r.t. the upper frontier of the class. NCS is a variant of the ELECTRE TRI method. We describe algorithms designed for learning such a model on the basis of assignment examples. We compare the performance of these algorithms and show show how they compare to other state of the art sorting methods methods (Choquistic Regression, UTADIS,…). We emphasize the interpretability of the NCS model and show how classification results can be explained.