Editor's Note
149 Italian saffron samples produced in the years from 2013 to 2016 in distinct sites located in five different Italian regions, Abruzzo, Tuscany, Umbria, Campania and Sardinia, together with 27 commercial samples, have been analyzed by high-performance liquid chromatography with diode array detection (HPLC-DAD). In this present study, two different analytical approaches aimed at distinguishing PDO saffron of L'Aquila from the other samples have been compared. The first strategy is a more traditional approach, where the chromatograms collected at specific wavelengths are classified by Partial Least Squares Discriminant Analysis (PLS-DA). The second strategy exploits the multi-way nature of data, avoiding discarding any source of information. Both approaches provided satisfactory predictions; the best results from the prediction point of view (estimated on an external set of samples) were achieved by the proposed multi-way methodology.
One hundred and forty-nine (149) Italian saffron samples produced in the years from 2013 to 2016 in distinct sites located in five different Italian regions, Abruzzo, Tuscany, Umbria, Campania and Sardinia, together with twenty-seven (27) commercial samples, have been analyzed by high-performance liquid chromatography with diode array detection (HPLC-DAD). Among the investigated samples, those produced in Abruzzo (L'Aquila area) present an even higher added-value, because, since 2005, saffron of L'Aquila has been granted of the protected designation of origin (PDO) mark. In the present study, two different analytical approaches aimed at distinguishing PDO saffron of L'Aquila from the other samples have been compared. The first strategy is a more traditional approach, where the chromatograms collected at specific wavelengths are classified by Partial Least Squares Discriminant Analysis (PLS-DA). The second strategy exploits the multi-way nature of data, avoiding discarding any source of information. Consequently, the entire spectro-chromatogram is handled by N-Partial Least Squares (N-PLS) and then classified by Linear Discriminant Analysis (LDA). Both approaches provided satisfactory predictions; the best results from the prediction point of view (estimated on an external set of samples) were achieved by the proposed multi-way methodology.

2020 Annual Fund Sponsors: