Pré-Publication, Document De Travail Année : 2025

Probabilistic models for permutations and dependence

Résumé

In order to improve the financial performance of a company, we must classify all their commercialized products according to their interests using financial criteria. The criteria and the final classification can be modeled using permutations. We consider Mallows' models defined in the space of permutations. We are particularly interested in the question of dependence in Mallows' models. Here, we introduce new machine learning approaches based on a cost function minimizing the impact of the dependence between criteria. In the multi criteria aggregation model, we consider some of the criteria as a permutation generated using a Mallows' model, in which the modal permutation is the permutation we want to find, while some other criteria are simulated using other criteria. Finally, the methodology is illustrated with a simulation study that compares the performances of the approaches.
Fichier principal
Vignette du fichier
probablistic_model_for_permutation_and_dependance.pdf (5) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04925265 , version 1 (01-02-2025)

Identifiants

  • HAL Id : hal-04925265 , version 1

Citer

Arthur Fétiveau, Gilles Durrieu, Emmanuel Frénod. Probabilistic models for permutations and dependence. 2025. ⟨hal-04925265⟩
0 Consultations
0 Téléchargements

Partager

More