Experimental perspectives on learning from imbalanced data

TLDR

When classes are imbalanced, many learning algorithms suffer reduced performance. The study investigates whether data sampling can improve learner performance on imbalanced data, how effectiveness depends on learner type, and whether results vary across performance metrics. The authors conduct a comprehensive suite of experiments on learning from imbalanced data. The results show that sampling frequently improves classifier performance.

Abstract

We present a comprehensive suite of experimentation on the subject of learning from imbalanced data. When classes are imbalanced, many learning algorithms can suffer from the perspective of reduced performance. Can data sampling be used to improve the performance of learners built from imbalanced data? Is the effectiveness of sampling related to the type of learner? Do the results change if the objective is to optimize different performance metrics? We address these and other issues in this work, showing that sampling in many cases will improve classifier performance.