Mining Negative and Positive Influence Rules Using Kullback-Leibler Divergence
Leïla Nemmiche Alachaher, Sylvie Guillaume · 2007
This paper describes a new method for mining negative and positive quantitative influence rules based on a coordination between a statistical dissimilarity measure (Kullback Leibler divergence) and contingency tables. This coordination identifies the significant positive and negative correlations and enables pertinent influence rules extraction.