International audienceLearning the parameters of a Majority Rule Sorting model (MR-Sort) through linear programming requires to use binary variables. In the context of preference learning where large sets of alternatives and numerous attributes are involved, such an approach is not an option in view of the large computing times implied. Therefore, we propose a new metaheuristic designed to learn the parameters of an MR-Sort model. This algorithm works in two phases that are iterated. The first one consists in solving a linear program determining the weights and the majority threshold, assuming a given set of profiles. The second phase runs a metaheuristic which determines profiles for a fixed set of weights and a majority threshold. The presentation focuses on the metaheuristic and reports the results of numerical tests, providing insights on the algorithm behavior

No 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 multicriteria decision analysis (MCDA), called MR‐Sort. This model determines the assigned class on the basis of a majority rule and an artificial object that is a typical lower profile of the category. MR‐Sort is a simplified variant of the ELECTRE TRI method. We describe an algorithm designed for learning such a model on the basis of assignment examples. We compare its performance with choquistic regression, a method recently proposed in the preference learning community, and with UTADIS, another MCDA method leaning on an additive value function (utility) model. Our experimentation shows that MR‐Sort competes with the other two methods, and leads to a model that is interpretable.