Cost-sensitive ensemble learning: a unifying framework

Abstract Over the years, a plethora of cost-sensitive methods have been proposed for learning on data when different types misclassification errors incur costs. Our contribution is unifying framework that provides comprehensive and insightful overview ensemble methods, pinpointing their differences similarities via fine-grained categorization. contains natural extensions generalisations ideas across be it AdaBoost, Bagging or Random Forest, as result not only yields all known to date but also some previously considered.