RESUMEN
Thermal properties of extra virgin olive oil (EVOO) consumed in China were investigated through differential scanning calorimetry (DSC), Principal Component Analysis (PCA) dimension reduction analysis combined with K-nearest neighbors (KNN) and support vector machine (SVM) models were used to determine whether oil samples would be EVOO. Both models exhibited a 100% ability to distinguish EVOO from non-EVOO. Additionally, the discrimination test of canola oil was still limited from EVOO adulteration due to its similarity of cooling profiles with EVOO. Canola oil was artificially added into EVOO in 5%, 10%, 20% and 30% and DSC test results collected were detected in computer models. Both models effectively identified even 5% canola oil adulterated in EVOO; however, KNN (93.75%) model exhibited much higher accuracy than SVM (43.92%). In conclusion, DSC combined with KMN model analysis was more available for the adulteration detection of EVOO.