/** * A predictive model wrapper and related classes that can be used to improve performance on highly * <a href="https://archive.nyu.edu/bitstream/2451/27763/2/CPP-02-00.pdf">imbalanced</a> datasets * with two possible classifications (a majority classification and a minority classification). * It works by reducing the imbalance by throwing away, at random, a proportion of the instances * with the majority classification. It then statistically corrects for this at prediction time. */ package quickml.supervised.classifier.downsampling;