A New Under-Sampling Method to Face Class Overlap and Imbalance

Class overlap and class imbalance are two data complexities that challenge the design of effective classifiers in Pattern Recognition and Data Mining as they may cause a significant loss in performance. Several solutions have been proposed to face both data difficulties, but most of these approaches...

Descripción completa

Detalles Bibliográficos
Autores: Guzmán Ponce, Angélica, Valdovinos Rosas, Rosa María, Sánchez Garreta, José Salvador, Marcial Romero, José Raymundo
Tipo de recurso: artículo
Fecha de publicación:2020
País:España
Institución:Universidad Católica San Antonio de Murcia (UCAM)
Repositorio:RIUCAM. Repositorio Institucional de la Universidad Católica San Antonio de Murcia
OAI Identifier:oai:repositorio.ucam.edu:10952/10769
Acceso en línea:http://hdl.handle.net/10952/10769
Access Level:acceso abierto
Palabra clave:Class imbalance
Class overlap
Under-sampling
Clustering
DBSCAN
Minimum spanning tree
id ES_6fe0855d3c8aaaac76dd7ae2dd7bfc2c
oai_identifier_str oai:repositorio.ucam.edu:10952/10769
network_acronym_str ES
network_name_str España
repository_id_str
spelling A New Under-Sampling Method to Face Class Overlap and ImbalanceGuzmán Ponce, AngélicaValdovinos Rosas, Rosa MaríaSánchez Garreta, José SalvadorMarcial Romero, José RaymundoClass imbalanceClass overlapUnder-samplingClusteringDBSCANMinimum spanning treeClass overlap and class imbalance are two data complexities that challenge the design of effective classifiers in Pattern Recognition and Data Mining as they may cause a significant loss in performance. Several solutions have been proposed to face both data difficulties, but most of these approaches tackle each problem separately. In this paper, we propose a two-stage under-sampling technique that combines the DBSCAN clustering algorithm to remove noisy samples and clean the decision boundary with a minimum spanning tree algorithm to face the class imbalance, thus handling class overlap and imbalance simultaneously with the aim of improving the performance of classifiers. An extensive experimental study shows a significantly better behavior of the new algorithm as compared to 12 state-of-the-art under-sampling methods using three standard classification models (nearest neighbor rule, J48 decision tree, and support vector machine with a linear kernel) on both real-life and synthetic databases.Ingeniería, Industria y ConstrucciónEscuela Politécnica2020info:eu-repo/semantics/articlehttp://hdl.handle.net/10952/10769reponame:RIUCAM. Repositorio Institucional de la Universidad Católica San Antonio de Murciainstname:Universidad Católica San Antonio de Murcia (UCAM)Inglésinfo:eu-repo/semantics/openAccessoai:repositorio.ucam.edu:10952/107692026-06-07T18:35:21Z
dc.title.none.fl_str_mv A New Under-Sampling Method to Face Class Overlap and Imbalance
title A New Under-Sampling Method to Face Class Overlap and Imbalance
spellingShingle A New Under-Sampling Method to Face Class Overlap and Imbalance
Guzmán Ponce, Angélica
Class imbalance
Class overlap
Under-sampling
Clustering
DBSCAN
Minimum spanning tree
title_short A New Under-Sampling Method to Face Class Overlap and Imbalance
title_full A New Under-Sampling Method to Face Class Overlap and Imbalance
title_fullStr A New Under-Sampling Method to Face Class Overlap and Imbalance
title_full_unstemmed A New Under-Sampling Method to Face Class Overlap and Imbalance
title_sort A New Under-Sampling Method to Face Class Overlap and Imbalance
dc.creator.none.fl_str_mv Guzmán Ponce, Angélica
Valdovinos Rosas, Rosa María
Sánchez Garreta, José Salvador
Marcial Romero, José Raymundo
author Guzmán Ponce, Angélica
author_facet Guzmán Ponce, Angélica
Valdovinos Rosas, Rosa María
Sánchez Garreta, José Salvador
Marcial Romero, José Raymundo
author_role author
author2 Valdovinos Rosas, Rosa María
Sánchez Garreta, José Salvador
Marcial Romero, José Raymundo
author2_role author
author
author
dc.subject.none.fl_str_mv Class imbalance
Class overlap
Under-sampling
Clustering
DBSCAN
Minimum spanning tree
topic Class imbalance
Class overlap
Under-sampling
Clustering
DBSCAN
Minimum spanning tree
description Class overlap and class imbalance are two data complexities that challenge the design of effective classifiers in Pattern Recognition and Data Mining as they may cause a significant loss in performance. Several solutions have been proposed to face both data difficulties, but most of these approaches tackle each problem separately. In this paper, we propose a two-stage under-sampling technique that combines the DBSCAN clustering algorithm to remove noisy samples and clean the decision boundary with a minimum spanning tree algorithm to face the class imbalance, thus handling class overlap and imbalance simultaneously with the aim of improving the performance of classifiers. An extensive experimental study shows a significantly better behavior of the new algorithm as compared to 12 state-of-the-art under-sampling methods using three standard classification models (nearest neighbor rule, J48 decision tree, and support vector machine with a linear kernel) on both real-life and synthetic databases.
publishDate 2020
dc.date.none.fl_str_mv 2020
dc.type.none.fl_str_mv info:eu-repo/semantics/article
format article
dc.identifier.none.fl_str_mv http://hdl.handle.net/10952/10769
url http://hdl.handle.net/10952/10769
dc.language.none.fl_str_mv Inglés
language_invalid_str_mv Inglés
dc.rights.none.fl_str_mv info:eu-repo/semantics/openAccess
eu_rights_str_mv openAccess
dc.source.none.fl_str_mv reponame:RIUCAM. Repositorio Institucional de la Universidad Católica San Antonio de Murcia
instname:Universidad Católica San Antonio de Murcia (UCAM)
instname_str Universidad Católica San Antonio de Murcia (UCAM)
reponame_str RIUCAM. Repositorio Institucional de la Universidad Católica San Antonio de Murcia
collection RIUCAM. Repositorio Institucional de la Universidad Católica San Antonio de Murcia
repository.name.fl_str_mv
repository.mail.fl_str_mv
_version_ 1869410539536384000
score 15,812455

Similares en LA Referencia