Imbalanced Classification: A Paradigm-Based Review

By Yang F. et al

Published on July 1, 2021

Read the original document by opening this link in a new tab.

Table of Contents

Abstract
Introduction
Three Classification Paradigms
A Matrix of Algorithms for Imbalanced Classification

Summary

This paper provides a paradigm-based review of imbalanced classification. It discusses the common issue of imbalanced classes in classification tasks and the challenges it poses. The paper explores different resampling techniques and paradigms for binary classification under imbalanced class sizes. It also investigates the combination of resampling techniques and classification methods through simulation studies and real data analysis. The findings demonstrate the complexities and interactions among resampling techniques, base classification methods, evaluation metrics, and imbalance ratios. The paper aims to provide guidance on choosing resampling techniques and base classification methods for practitioners dealing with imbalanced classification tasks.