An Experimental Study of Different Ordinal Regression Methods and Measures

Áreas de investigación:

• Methodology - Machine learning

Año:

2012

Tipo de publicación:

Artículo en conferencia

Palabras clave:

Ordinal classification, ordinal regression, evaluation measures, accuracy, support vector machines, threshold models, ordinal logistic regression

Autores:

• Gutiérrez, Pedro Antonio
• Pérez-Ortiz, María
• Fernandez-Navarro, Francisco
• Sánchez-Monedero, Javier
• Hervás-Martínez, César

Título del libro:

7th International Conference on Hybrid Artificial Intelligence Systems (HAIS2012)

Páginas:

296-307

BibTex:

@conference{ordinalReview2012, author = "Pedro Antonio Guti{\'e}rrez and Mar{\'i}a P{\'e}rez-Ortiz and Francisco Fernandez-Navarro and Javier S{\'a}nchez-Monedero and C{\'e}sar Herv{\'a}s-Mart{\'i}nez", abstract = "In this paper, an experimental study of different ordinal regression methods and measures is presented. The first objective is to gather the results of a considerably high number of methods, datasets and measures, since there are not many previous comparative studies of this kind in the literature. The second objective is to detect the redundancy between the evaluation measures used for ordinal regression. The results obtained present the maximum MAE (maximum of the mean absolute error of the difference between the true and the predicted ranks of the worst classified class) as a very interesting alternative for ordinal regression, being the less uncorrelated with respect to the rest of measures. Additionally, SVOREX and SVORIM are found to yield very good performance when the objective is to minimize this maximum MAE. ", booktitle = "7th International Conference on Hybrid Artificial Intelligence Systems (HAIS2012)", doi = "10.1007/978-3-642-28931-6_29", keywords = "Ordinal classification, ordinal regression, evaluation measures, accuracy, support vector machines, threshold models, ordinal logistic regression", pages = "296-307", title = "{A}n {E}xperimental {S}tudy of {D}ifferent {O}rdinal {R}egression {M}ethods and {M}easures", url = "http://dx.doi.org/10.1007/978-3-642-28931-6_29", year = "2012", }

Abstract:

In this paper, an experimental study of different ordinal regression methods and measures is presented. The first objective is to gather the results of a considerably high number of methods, datasets and measures, since there are not many previous comparative studies of this kind in the literature. The second objective is to detect the redundancy between the evaluation measures used for ordinal regression. The results obtained present the maximum MAE (maximum of the mean absolute error of the difference between the true and the predicted ranks of the worst classified class) as a very interesting alternative for ordinal regression, being the less uncorrelated with respect to the rest of measures. Additionally, SVOREX and SVORIM are found to yield very good performance when the objective is to minimize this maximum MAE.