A variable selection method for multiclass classification problems using two-class ROC analysis
2018
de Figueiredo, Miguel | Cordella, Christophe | Jouan-Rimbaud Bouveresse, Delphine | Archer, Xavier | Bégué, Jean-Marc | Rutledge, Douglas | Ingénierie, Procédés, Aliments (GENIAL) ; Institut National de la Recherche Agronomique (INRA)-AgroParisTech | Université Paris-Saclay | Laboratoire Central de Préfecture de Police ; Laboratoire Central de Préfecture de Police | Physiologie de la Nutrition et du Comportement Alimentaire (PNCA) ; Institut National de la Recherche Agronomique (INRA)-AgroParisTech | Laboratoire Central de la préfecture de Police de Paris
Modern procedures in analytical chemistry generate enormous amounts of data, which must beprocessed and interpreted. The treatment of such high-dimensional datasets often necessitatesthe prior selection of a reduced number of variables in order to extract knowledge about thesystem under study and to maximize the predictability of the models built. Therefore, thisarticle describes a variable selection method for multiclass classification problems using two-classROC analysis and its associated area under the ROC curve as a variable selection criterion.The variable selection method has been successfully applied to two datasets. For comparisonpurposes, two other variable selection methods, ReliefF and mRMR, were used and double cross-validation PLS-DA was applied using: (1) all variables and (2) the variables selected using thethree methods. It has been demonstrated that correct variable selection can substantially reducethe dimensionality of the datasets, while maximizing the predictability of the models.
Показать больше [+] Меньше [-]Ключевые слова АГРОВОК
Библиографическая информация
Эту запись предоставил Institut national de la recherche agronomique