Discovering interesting prediction rules with a genetic algorithm

In essence, the goal of data mining is to discover knowledge which is highly accurate, comprehensible and “interesting” (surprising, novel). Although the literature emphasizes predictive accuracy and comprehensibility, the discovery of interesting knowledge remains a formidable challenge for data mining algorithms. We present a genetic algorithm designed from the scratch to discover interesting rules. Our GA addresses the dependence modelling task, where different rules can predict different goal attributes. This task can be regarded as a generalization of the classification task, where all rules predict the same goal attribute