On the Importance of Diversity in Re-Sampling for Imbalanced Data and Rare Events in Mortality Risk Models

Yang, Yuxuan; Khorshidi, Hadi Akbarzadeh; Aickelin, Uwe; Nevgi, Aditi; Ekinci, Elif

doi:10.1145/3437378.3437382

Cited by 3 publications

(1 citation statement)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In 2021, a diversity-based sampling method with a drop-in functionality was proposed to evaluate diversity. It was achieved via a greedy algorithm that is used to identify and discard subsets that share the most similarity [12]. KMEANS-SMOTE [13] is a data-level oversampling method that was introduced in 2018 which combines k-means clustering algorithm with SMOTE.…”

Section: Related Workmentioning

confidence: 99%

A Diversity-Based Synthetic Oversampling Using Clustering for Handling Extreme Imbalance

Yang,

Akbarzadeh Khorshidi,

Aickelin

2023

SN COMPUT. SCI.

View full text Add to dashboard Cite

Imbalanced data are typically observed in many real-life classification problems. However, mainstream machine learning algorithms are mostly designed with the underlying assumption of a relatively well-balanced distribution of classes. The mismatch between reality and algorithm assumption results in a deterioration of classification performance. One form of approach to address this problem is through re-sampling methods, although its effectiveness is limited; most re-sampling methods fail to consider the distribution of minority and majority instances and the diversity within synthetically generated data. Diversity becomes increasingly important when minority data becomes more sparse, as each data point becomes more valuable. They should all be considered during the generation process instead of being regarded as noise. In this paper, we propose a cluster-based diversity re-sampling method, combined with NOAH algorithm. Neighbourhood-based Clustering Diversity Over-sampling (NBCDO) is introduced with the aim to complement our previous cluster-based diversity algorithm Density-based Clustering Diversity Over-sampling (DBCDO). It first uses a neighbourhood-based clustering algorithm to consider the distribution of both minority and majority class instances, before applying NOAH algorithm to encourage diversity optimisation during the generation of synthetic instances. We demonstrate the implementation of both cluster-based diversity methods by conducting experiments over 10 real-life datasets with ≤ 5% imbalance ratio and show that our proposed cluster-based diversity algorithm (NBCDO, DBCDO) brings performance improvements over its comparable methods (DB-SMOTE, MAHAKIL, KMEANS-SMOTE, MC-SMOTE).

show abstract

Section: Related Workmentioning

confidence: 99%

A Diversity-Based Synthetic Oversampling Using Clustering for Handling Extreme Imbalance

Yang,

Akbarzadeh Khorshidi,

Aickelin

2023

SN COMPUT. SCI.

View full text Add to dashboard Cite

show abstract

AutoScore-Imbalance: An interpretable machine learning tool for development of clinical scores with rare events data

Yuan

Xie

Ong

et al. 2022

Journal of Biomedical Informatics

View full text Add to dashboard Cite

A review on over-sampling techniques in classification of multi-class imbalanced datasets: insights for medical problems

Yang,

Khorshidi,

Aickelin

2024

Front. Digit. Health

View full text Add to dashboard Cite

There has been growing attention to multi-class classification problems, particularly those challenges of imbalanced class distributions. To address these challenges, various strategies, including data-level re-sampling treatment and ensemble methods, have been introduced to bolster the performance of predictive models and Artificial Intelligence (AI) algorithms in scenarios where excessive level of imbalance is present. While most research and algorithm development have been focused on binary classification problems, in health informatics there is an increased interest in the field to address the problem of multi-class classification in imbalanced datasets. Multi-class imbalance problems bring forth more complex challenges, as a delicate approach is required to generate synthetic data and simultaneously maintain the relationship between the multiple classes. The aim of this review paper is to examine over-sampling methods tailored for medical and other datasets with multi-class imbalance. Out of 2,076 peer-reviewed papers identified through searches, 197 eligible papers were chosen and thoroughly reviewed for inclusion, narrowing to 37 studies being selected for in-depth analysis. These studies are categorised into four categories: metric, adaptive, structure-based, and hybrid approaches. The most significant finding is the emerging trend toward hybrid resampling methods that combine the strengths of various techniques to effectively address the problem of imbalanced data. This paper provides an extensive analysis of each selected study, discusses their findings, and outlines directions for future research.

show abstract

On the Importance of Diversity in Re-Sampling for Imbalanced Data and Rare Events in Mortality Risk Models

Cited by 3 publications

References 20 publications

A Diversity-Based Synthetic Oversampling Using Clustering for Handling Extreme Imbalance

A Diversity-Based Synthetic Oversampling Using Clustering for Handling Extreme Imbalance

AutoScore-Imbalance: An interpretable machine learning tool for development of clinical scores with rare events data

A review on over-sampling techniques in classification of multi-class imbalanced datasets: insights for medical problems

Contact Info

Product

Resources

About