A study on feature selection based on AICc and its application to microarray data

Abstract

The following paper contains a report of the nal project written at the Faculty of Informatics (Facultat d'Inform atica de Barcelona) on Techni- cal University of Catalonia (Universitat Polit ecnica de Catalunya) under the supervision of professor Luis Belanche. The project provides R lan- guage procedures, collected in a library, to select the best features, among the given feature set within the data, to describe a classi er. The prin- cipal criterion in such a selection process would be Akaike's information criterion - AIC.