Qualité d'un graphe implicatif : variance

Abstract : A weighted graph, without cycle, is a representation of a set of implicative association
rules extracted from a digital cross variables and subjects. The problem of homogeneity,
consistency and therefore the relevance of the interpretations of expert arises since in
Statistical Implicative Analysis (SIA) it is possible to vary the threshold of representation of
partial rules. We present here the concept of variance implicative like the classical concept of
variance in order to characterize the homogeneity of the representation. It relies on a
metaphor of repulsion versus mutual consistency between two binary variables from their
symmetric difference.
Complete list of metadatas

Contributor : Jean-Claude Regnier <>
Submitted on : Tuesday, June 16, 2009 - 6:28:34 PM
Last modification on : Friday, June 14, 2019 - 1:20:53 AM


  • HAL Id : halshs-00396034, version 1


Régis Gras, Jean-Claude Regnier. Qualité d'un graphe implicatif : variance
implicative. Régis Gras; Jean-Claude Régnier; Fabrice Guillet. Analyse statistique implicative: Une méthode d'analyse de données pour la recherche de causalités., Éditions Cépaduès, pp.151 -163, 2009, RNTI-E-16. ⟨halshs-00396034⟩



Record views