Barbieri, SaraCevoli, ChiaraBendini, AlessandraQuintanilla-Casas, BeatrizGarcía González, Diego LuísGallina Toschi, Tullia2021-02-092021-02-092020-07-022304-8158https://hdl.handle.net/2445/173746This research aims to develop a classification model based on untargeted elaboration of volatile fraction fingerprints of virgin olive oils (n = 331) analyzed by flash gas chromatography to predict the commercial category of samples (extra virgin olive oil, EVOO; virgin olive oil, VOO; lampante olive oil, LOO). The raw data related to volatile profiles were considered as independent variables, while the quality grades provided by sensory assessment were defined as a reference parameter. This data matrix was elaborated using the linear technique partial least squares-discriminant analysis (PLS-DA), applying, in sequence, two sequential classification models with two categories (EVOO vs. no-EVOO followed by VOO vs. LOO and LOO vs. no-LOO followed by VOO vs. EVOO). The results from this large set of samples provide satisfactory percentages of correctly classified samples, ranging from 72% to 85%, in external validation. This confirms the reliability of this approach in rapid screening of quality grades and that it represents a valid solution for supporting sensory panels, increasing the effciency of the controls, and also applicable to the industrial sector.11 p.application/pdfengcc-by (c) Barbieri, Sara et al., 2020http://creativecommons.org/licenses/by/3.0/esOli d'olivaCromatografia de gasosTraçabilitatOlive oilGas chromatographyTraceabilityinfo:eu-repo/semantics/article7018362021-02-09info:eu-repo/semantics/openAccess32630810