Measuring water residue in olive oil by means of a smartphone-connected pocket spectrometer and artificial intelligence

摘要

SCiO is a smartphone-connected pocket spectrometer operating in the 700-1100 nm band. Together with a learning machine algorithm, it already demonstrated the effectiveness for distinguishing extra virgin from non extra virgin olive oils and for the multi-analysis of nutraceutical indicators. This paper shows a new experiment for the assessment of water residue at the end of the olive oil production process. Principal Component Analysis and Linear Discriminant Analysis were used to demonstrate a qualitative screening with a threshold of 0.5% v/v of water content and an accuracy of 93%. Also, a model for predicting the water concentration was created by means of the Partial Least Square regression, providing a regression coefficient R~2 =0.92, and an error of 0.26%.