The most representative utility function for non-additive robust ordinal regression

S. Angilella, Salvatore Greco, B. Matarazzo

    Research output: Chapter in Book/Report/Conference proceedingChapter (peer-reviewed)peer-review

    Abstract

    Non-additive robust ordinal regression (NAROR) considers Choquet integral or one of its generalizations to represent preferences of a Decision Maker (DM). More precisely, NAROR takes into account all the fuzzy measures which are compatible with the preference information given by the DM and builds two preference relations: possible preference relation, when there is at least one compatible fuzzy measure for which an alternative is preferred to the other, and necessary preference relation, when an alternative is preferred to the other for all compatible fuzzy measures. Although it is interesting to take into consideration all the compatible fuzzy measures, in some decision problems we need to give a value to every alternative and it results necessary to obtain the most representative fuzzy measures among all the compatible ones. The aim of the paper is to propose an algorithm to the DM for selecting the most representative utility function expressed as Choquet integral from which a DM’s representation of preferences is obtained.
    Original languageEnglish
    Title of host publicationComputational intelligence for knowledge-based systems design: 13th international conference on information processing and management of uncertainty
    EditorsE. Hullermeier, R. Kruse, F. Hoffmann
    Place of PublicationBerlin
    PublisherSpringer
    Pages220-229
    Number of pages10
    Volume6178
    Edition6178
    ISBN (Print)9783642140488
    DOIs
    Publication statusPublished - Jul 2010

    Publication series

    NameLecture notes in computer science
    PublisherSpringer
    Number6178
    ISSN (Print)0302-9743

    Fingerprint

    Dive into the research topics of 'The most representative utility function for non-additive robust ordinal regression'. Together they form a unique fingerprint.

    Cite this