Cluster validity indices for automatic clustering: A comprehensive review

The Cluster Validity Index is an integral part of clustering algorithms. It evaluates inter-cluster separation and intra-cluster cohesion of candidate clusters to determine the quality of potential solutions. Several cluster validity indices have been suggested for both classical clustering algorith...

Full description

Saved in:
Bibliographic Details
Main Authors: Abiodun M. Ikotun, Faustin Habyarimana, Absalom E. Ezugwu
Format: Article
Language:English
Published: Elsevier 2025-01-01
Series:Heliyon
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2405844025003330
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1832573068911837184
author Abiodun M. Ikotun
Faustin Habyarimana
Absalom E. Ezugwu
author_facet Abiodun M. Ikotun
Faustin Habyarimana
Absalom E. Ezugwu
author_sort Abiodun M. Ikotun
collection DOAJ
description The Cluster Validity Index is an integral part of clustering algorithms. It evaluates inter-cluster separation and intra-cluster cohesion of candidate clusters to determine the quality of potential solutions. Several cluster validity indices have been suggested for both classical clustering algorithms and automatic metaheuristic-based clustering algorithms. Different cluster validity indices exhibit different characteristics based on the mathematical models they employ in determining the values for the various cluster attributes. Metaheuristic-based automatic clustering algorithms use cluster validity index as a fitness function in its optimization procedure to evaluate the candidate cluster solution's quality. A systematic review of the cluster validity indices used as fitness functions in metaheuristic-based automatic clustering algorithms is presented in this study. Identifying, reporting, and analysing various cluster validity indices is important in classifying the best CVIs for optimum performance of a metaheuristic-based automatic clustering algorithm. This review also includes an experimental study on the performance of some common cluster validity indices on some synthetic datasets with varied characteristics as well as real-life datasets using the SOSK-means automatic clustering algorithm. This review aims to assist researchers in identifying and selecting the most suitable cluster validity indices (CVIs) for their specific application areas.
format Article
id doaj-art-3c9f6e03bdee44d9be9645292a9ca4b7
institution Kabale University
issn 2405-8440
language English
publishDate 2025-01-01
publisher Elsevier
record_format Article
series Heliyon
spelling doaj-art-3c9f6e03bdee44d9be9645292a9ca4b72025-02-02T05:28:38ZengElsevierHeliyon2405-84402025-01-01112e41953Cluster validity indices for automatic clustering: A comprehensive reviewAbiodun M. Ikotun0Faustin Habyarimana1Absalom E. Ezugwu2School of Mathematics, Statistics, and Computer Science, University of KwaZulu-Natal, King Edward Avenue, Pietermaritzburg Campus, Pietermaritzburg, 3201, KwaZulu-Natal, South AfricaSchool of Mathematics, Statistics, and Computer Science, University of KwaZulu-Natal, King Edward Avenue, Pietermaritzburg Campus, Pietermaritzburg, 3201, KwaZulu-Natal, South AfricaUnit for Data Science and Computing, North-West University, 11 Hoffman Street, Potchefstroom, 2520, North-West, South Africa; Corresponding author.The Cluster Validity Index is an integral part of clustering algorithms. It evaluates inter-cluster separation and intra-cluster cohesion of candidate clusters to determine the quality of potential solutions. Several cluster validity indices have been suggested for both classical clustering algorithms and automatic metaheuristic-based clustering algorithms. Different cluster validity indices exhibit different characteristics based on the mathematical models they employ in determining the values for the various cluster attributes. Metaheuristic-based automatic clustering algorithms use cluster validity index as a fitness function in its optimization procedure to evaluate the candidate cluster solution's quality. A systematic review of the cluster validity indices used as fitness functions in metaheuristic-based automatic clustering algorithms is presented in this study. Identifying, reporting, and analysing various cluster validity indices is important in classifying the best CVIs for optimum performance of a metaheuristic-based automatic clustering algorithm. This review also includes an experimental study on the performance of some common cluster validity indices on some synthetic datasets with varied characteristics as well as real-life datasets using the SOSK-means automatic clustering algorithm. This review aims to assist researchers in identifying and selecting the most suitable cluster validity indices (CVIs) for their specific application areas.http://www.sciencedirect.com/science/article/pii/S2405844025003330ClusteringCluster validity indexAutomatic clusteringMetaheuristic algorithmsOptimization algorithms
spellingShingle Abiodun M. Ikotun
Faustin Habyarimana
Absalom E. Ezugwu
Cluster validity indices for automatic clustering: A comprehensive review
Heliyon
Clustering
Cluster validity index
Automatic clustering
Metaheuristic algorithms
Optimization algorithms
title Cluster validity indices for automatic clustering: A comprehensive review
title_full Cluster validity indices for automatic clustering: A comprehensive review
title_fullStr Cluster validity indices for automatic clustering: A comprehensive review
title_full_unstemmed Cluster validity indices for automatic clustering: A comprehensive review
title_short Cluster validity indices for automatic clustering: A comprehensive review
title_sort cluster validity indices for automatic clustering a comprehensive review
topic Clustering
Cluster validity index
Automatic clustering
Metaheuristic algorithms
Optimization algorithms
url http://www.sciencedirect.com/science/article/pii/S2405844025003330
work_keys_str_mv AT abiodunmikotun clustervalidityindicesforautomaticclusteringacomprehensivereview
AT faustinhabyarimana clustervalidityindicesforautomaticclusteringacomprehensivereview
AT absalomeezugwu clustervalidityindicesforautomaticclusteringacomprehensivereview