FTIR coupled with machine learning to unveil spectroscopic benchmarks in the Italian EVOO

Non‐destructive analytical analyses coupled with classification and regression algorithms are promising techniques for monitoring quality, traceability and safety assessment in food industry. To prevent food fraud, Italian Extra Virgin Olive Oil (EVOO) is particularly held in check. Here, attenuated total Reflectance‐Fourier Transform Infrared spectroscopy (ATR‐FTIR) with Machine Learning is carried out to study an Italian EVOO data set coming from 6 regions to verify the geographical traceability, the cultivar and the repeatability of the agronomical practices, till the adulterated EVOO from soy and corn. The present work is carried out without reagents or esterification processes and considering the entire frequency range without any spectral windows selection, drastically reducing time and costs. Toscana, Lazio, Puglia and Calabria result regions well reproducible in terms of geo‐traceability, unlike the Sicilia and Umbria. The model extracts spectral benchmarks in EVOO in the following vibrational modes at 3004, 2952, 2922, 2852, 1742 and 1160 cm‐1. Spectroscopic fingerprints and chemical descriptors to discriminate a wide Italian EVOOs. Non‐destructive innovative FTIR‐ML methodology to execute fast and reliable quality checks in the EVOO discrimination. Cost and time‐consuming reduction with no spectral windows selection. SIMCA, ν‐SVMR and PCR models to benchmark Italian EVOO.