Coupling REPMAC with FDA to solve highly imbalanced classification problems

In many critical real world classification problems one of the classes has much less samples than the others (class imbalance). In a previous work we introduced the REPMAC algorithm to solve imbalanced problems. Using a clustering method, REPMAC recursively splits the majority class in several subse...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Ahumada, Hernán César, Grinblat, Guillermo L., Uzal, Lucas, Ceccatto, Hermenegildo Alejandro, Granitto, Pablo Miguel
Formato: Objeto de conferencia
Lenguaje:Inglés
Publicado: 2008
Materias:
Ciencias Informáticas
imbalanced problems
Algorithms
Acceso en línea:http://sedici.unlp.edu.ar/handle/10915/21686
Aporte de:
SEDICI (UNLP) de Universidad Nacional de La Plata
id I19-R120-10915-21686
record_format dspace
institution Universidad Nacional de La Plata
institution_str I-19
repository_str R-120
collection SEDICI (UNLP)
language Inglés
topic Ciencias Informáticas
imbalanced problems
Algorithms
spellingShingle Ciencias Informáticas
imbalanced problems
Algorithms
Ahumada, Hernán César
Grinblat, Guillermo L.
Uzal, Lucas
Ceccatto, Hermenegildo Alejandro
Granitto, Pablo Miguel
Coupling REPMAC with FDA to solve highly imbalanced classification problems
topic_facet Ciencias Informáticas
imbalanced problems
Algorithms
description In many critical real world classification problems one of the classes has much less samples than the others (class imbalance). In a previous work we introduced the REPMAC algorithm to solve imbalanced problems. Using a clustering method, REPMAC recursively splits the majority class in several subsets, creating a decision tree, until the resulting sub-problems are balanced or easy to solve. In this work we evaluate the use of three different classifiers coupled with REPMAC. We compare the perfomance of those methods using 7 datasets from the UCI repository spanning a wide range of number of features and imbalance degree. We find that the good perfomance of REPMAC is almost independent of the classifier coupled to it, which suggest that it success is mostly related to the use of an appropriate strategy to cope with imbalanced problems
format Objeto de conferencia
Objeto de conferencia
author Ahumada, Hernán César
Grinblat, Guillermo L.
Uzal, Lucas
Ceccatto, Hermenegildo Alejandro
Granitto, Pablo Miguel
author_facet Ahumada, Hernán César
Grinblat, Guillermo L.
Uzal, Lucas
Ceccatto, Hermenegildo Alejandro
Granitto, Pablo Miguel
author_sort Ahumada, Hernán César
title Coupling REPMAC with FDA to solve highly imbalanced classification problems
title_short Coupling REPMAC with FDA to solve highly imbalanced classification problems
title_full Coupling REPMAC with FDA to solve highly imbalanced classification problems
title_fullStr Coupling REPMAC with FDA to solve highly imbalanced classification problems
title_full_unstemmed Coupling REPMAC with FDA to solve highly imbalanced classification problems
title_sort coupling repmac with fda to solve highly imbalanced classification problems
publishDate 2008
url http://sedici.unlp.edu.ar/handle/10915/21686
work_keys_str_mv AT ahumadahernancesar couplingrepmacwithfdatosolvehighlyimbalancedclassificationproblems
AT grinblatguillermol couplingrepmacwithfdatosolvehighlyimbalancedclassificationproblems
AT uzallucas couplingrepmacwithfdatosolvehighlyimbalancedclassificationproblems
AT ceccattohermenegildoalejandro couplingrepmacwithfdatosolvehighlyimbalancedclassificationproblems
AT granittopablomiguel couplingrepmacwithfdatosolvehighlyimbalancedclassificationproblems
bdutipo_str Repositorios
_version_ 1764820464805347329

Ejemplares similares