A robustness study of calibration models for olive oil classification: Targeted and non-targeted fingerprint approaches based on GC-IMS.

The dual separation in gas chromatography-ion mobility spectrometry generates complex multi-dimensional data, whose interpretation is a challenge. In this work, two chemometric approaches for olive oil classification are compared to get the most robust model over time: i) an non-targeted fingerprinting analysis, in which the overall GC-IMS data was processed and ii) a targeted approach based on peak-region features (markers). A total of 701 olive samples from two harvests (2014-2015 and 2015-2016) were analysed and processed by both approaches. The models built with data samples of 2014-2015 showed that both approaches were suitable for samples classification (success >74%). However, when these models were applied for classifying samples from 2015-2016, better values were obtained using markers. The combination of data from the two harvests to build the chemometric models improved the percentages of success (>90%). These results confirm the potential of GC-IMS based approaches for olive oil classification.