Extension of the Statistical Implicative Analysis to Rule Hierarchies
Régis Gras, Antoine Bodin, Raphaël Couturier, Pablo Gregori · 2023
In this chapter, we extend the classical notion of quasi-implication to “rules of rules”—also called R-rules—, whose premises and conclusions can be eventually rules themselves. A new statistical measure, based on the implicative intensity is defined to assess the significance of the R-rules on a data set. We show how to organize these R-rules in a new combinatorial structure, the oriented hierarchy, which is inspired by the classical hierarchical classification. An incremental algorithm is developed to found the most significant R-rule. An illustration is presented on a real data set 1 .