RT info:eu-repo/semantics/article
T1 On feature selection protocols for very low-sample-size data
A1 Kuncheva, Ludmila I. .
A1 Rodríguez Diez, Juan José
K1 Feature selection
K1 Wide datasets
K1 Experimental protoco
K1 Training/testing
K1 Cross-validation
K1 Computer science
K1 Informática
AB High-dimensional data with very few instances are typical in many application domains. Selecting a highly discriminative subset of the original features is often the main interest of the end user. The widely-used feature selection protocol for such type of data consists of two steps. First, features are selected from the data (possibly through cross-validation), and, second, a cross-validation protocol is applied to test a classifier using the selected features. The selected feature set and the testing accuracy are then returned to the user. For the lack of a better option, the same low-sample-size dataset is used in both steps. Questioning the validity of this protocol, we carried out an experiment using 24 high-dimensional datasets, three feature selection methods and five classifier models. We found that the accuracy returned by the above protocol is heavily biased, and therefore propose an alternative protocol which avoids the contamination by including both steps in a single cross-validation loop. Statistical tests verify that the classification accuracy returned by the proper protocol is significantly closer to the true accuracy (estimated from an independent testing set) compared to that returned by the currently favoured protocol.
PB Elsevier
SN 0031-3203
YR 2018
FD 2018-09
LK http://hdl.handle.net/10259/4814
UL http://hdl.handle.net/10259/4814
LA eng
NO project RPG-2015-188 funded by The Leverhulme Trust, UK and by project TIN2015-67534-P (MINECO/FEDER, UE) funded by the Ministerio de Economía y Competitividad of the Spanish Government and European Union FEDER funds
DS Repositorio Institucional de la Universidad de Burgos
RD 23-nov-2024