Exploring the behaviour of base classifiers in credit scoring ensembles
Scholar | Otros documentos del autor: Marqués Marzal, Ana Isabel; García Jiménez, Vicente; Sánchez Garreta, José Salvador
MetadatosMostrar el registro completo del ítem
TítuloExploring the behaviour of base classifiers in credit scoring ensembles
Fecha de publicación2012
Versión del editorhttp://ac.els-cdn.com/S0957417412003363/1-s2.0-S0957417412003363-main.pdf?_tid=e ...
Many techniques have been proposed for credit risk assessment, from statistical models to artificial intelligence methods. During the last few years, different approaches to classifier ensembles have successfully been ... [+]
Many techniques have been proposed for credit risk assessment, from statistical models to artificial intelligence methods. During the last few years, different approaches to classifier ensembles have successfully been applied to credit scoring problems, demonstrating to be more accurate than single prediction models. However, it is still a question what base classifiers should be employed in each ensemble in order to achieve the highest performance. Accordingly, the present paper evaluates the performance of seven individual prediction techniques when used as members of five different ensemble methods. The ultimate aim of this study is to suggest appropriate classifiers for each ensemble approach in the context of credit scoring. The experimental results and statistical tests show that the C4.5 decision tree constitutes the best solution for most ensemble methods, closely followed by the multilayer perceptron neural network and logistic regression, whereas the nearest neighbour and the naive Bayes classifiers appear to be significantly the worst. [-]
Tipo de documentoinfo:eu-repo/semantics/article
Derechos de acceso
© 2012 Elsevier Ltd. All rights reserved.
Aparece en las colecciones
- INIT_Articles