Laser-based classification of olive oils assisted by machine learning. (1st January 2020)

• Record Type: Journal Article
• Authors: Gazeli, Odhisea; Bellou, Elli; Stefas, Dimitrios; Couris, Stelios
• Is Part Of: Food chemistry. Volume 302(2019)
• Publication Date: 2020-01-01
• Subjects: Laser-induced breakdown spectroscopy (LIBS) -- Olive oil -- Acidity -- LDA, SVM and RFC algorithmic models -- Chemometrics; Food -- Analysis -- Periodicals; Food -- Composition -- Periodicals; 664
• DOI: 10.1016/j.foodchem.2019.125329
• Languages: English
• ISSNs: 0308-8146

Abstract

Highlights: LIBS spectra of different quality olive oils were obtained and analyzed. LIBS spectral features of olive oil correlated with oil acidity and geographical origin. PCA, LDA, SVM and RFC models classify oils based on acidity and designation of origin. Classification accuracies of the tested models ranged between 90 and 99%. Abstract: Olive oil is an essential diet component in all Mediterranean countries having a considerable impact on the local economies, which are producing almost 90% of the world production. Therefore, the quality assessment of olive oil in terms of its acidity and its authentication in terms of PDO (Protected Designation of Origin) and PGI (Protected Geographical Indications) characterizations are nowadays necessary and of great importance for the market of olive oil and the related economic activities. In the present work, Laser Induced Breakdown Spectroscopy (LIBS) is used assisted by machine learning algorithms for retrieving of the information contained in the LIBS spectra to provide a simple, reliable, and ultrafast methodology for olive oils classification in terms of the degree of acidity and geographical origin. The combination of LIBS technique with machine learning statistical analysis approaches constitute a very powerful tool for the fast, in-situ and remote quality control of olive oil.