Mining models of exceptional objects through rule learning

A new technique, SNIPER, is proposed for learning a model that deals with continuous values of exceptionality. Specifically, given some training objects associated with a continuous attribute F, SNIPER induces a rule-based model for the identification of those objects likely to score the maximum values for F. The purpose of SNIPER differs from the one pursued in regression problems, since its main objective is to retrieve those objects more likely to score the highest values of F. Although there are opportunities for improvement, the results of a preliminary evaluation are encouraging. SNIPER is competitive in the quality of the attained results with respect to some established competitors, while outperforming them when the exceptional objects are very rare. Additionally, SNIPER is much faster in the induction of a model of object exceptionality.