Adulteration detection in olive oil using dielectric technique and data mining

AbstractOlive oil is one of the most important agricultural crops due to its digestive properties and economic status. However, olive oil production is a costly process which causes an expensive price of the final product. The most jobbery ways during olive oil production consist of mixing other oils such as maize, sunflower and soya oil into the olive oil. So, the aim of this study was to develop a dielectric-based system to detect adulteration in olive oil using cylindrical capacitive sensor. For categorizing of fake olive oil by using frequency specification, Linear Discriminant Analysis (LDA) was developed. A set of 15 samples of olive oil, sunflower oil and canola oil which mixed with different ratio of adulteration, were used for calibration and evaluation of developed system. For each sample, 25 iterations were performed. Regarding results, the highest error rate was for a sample containing 60% olive oil-40% canola oil. In general, 7 iterations failed to be properly recognized. Regarding to accuracy indexes, specificity and sensitivity, the system had the minimum error for a mixed sample (60% olive oil-40% canola oil), specificity and sensitivity were obtained as 98% and 100%, respectively and accuracy was obtained as 72%, which was the weakest value. Finally, regarding mean value table for all sample, accuracy reached to 97%. Results showed the developed technique has a good capability of detecting impurities in olive oil. It is concluded from obtained results that the developed system revealed an acceptable adulterated detection in oil production