Modern analytical measurement technologies for food authentication, such as infrared, NMR, mass spectrometry and chromatography, produce the data (spectrum, chromatogram) recorded in digital form. A measurement on a single sample typically comprises thousands of numbers, which is many more than the number of samples, meaning that the experiment overall is underdetermined. Furthermore, chemically different specimens often give rise to quite similar measurements, especially in some of the spectroscopy methods, where there are large numbers of overlapped spectral bands. The techniques of multivariate analysis are especially suitable for dealing with this kind of data to get the best out of these complex and unwieldy datasets.
In this scientific opinion paper, the advantages of a multivariate strategy compared with univariate assessments are discussed, and selected techniques that are now well established in analytical chemistry, such as the data compression methods of principal component analysis are examined. Predictive approaches suitable for authentication applications: discriminant and classification strategies, and class-modelling techniques are also considered. Validation is critical to the application of multivariate techniques. Also, the wider aspects of experimental design, such as the importance of representative sampling are discussed and illustrated from real-world examples of food authenticity problems.
Read the abstract here