Nuclear magnetic resonance profiling, combined with a single-layer artificial neural network, is used for the evaluation of the content of mixtures of different kinds of milk. In particular, aqueous fractions of cow and sheep milk mixtures are analyzed by (1) H NMR. The spectral differences are highlighted by an analysis of the variance and a principal component analysis. The species classification problem is solved by a linear discriminant analysis. The quantification of the relative amount of the milk of two different species is then achieved by solving the appropriate multilinear problem.

Nuclear magnetic resonance (NMR) profiling is used for characterization of monocultivar binary wine mixtures. Classification and quantification of the relative amount of wine in the mixture are made in two steps. First, each sample is classified as a mixture of a determined type by solving the appropriate classification problem using NMR profiles. The relative amount of the two corresponding monovarietal wines is then evaluated by multilinear regression of a selected set of NMR variables. Linear discriminant analysis (LDA), used in the classification step, gives a very good separation among the different mixture classes. On the other hand, a single layer artificial neural network, used to solve the multilinear problem, gives the relative amount of wine type in the mixture with a precision of about 10%.

Proton NMR profiling is nowadays a consolidated technique for the identification of geographical origin of food samples. The common approach consists in correlating NMR spectra of food samples to their territorial origin by multivariate classification statistical algorithms. In the present work, we illustrate an alternative perspective to exploit territorial information, contained in the NMR spectra, which is based on the implementation of a geographic information system (GIS). Nuclear magnetic resonance spectra are used to build a GIS map permitting the identification of territorial regions having strong similarities in the chemical content of the produced food (terroir units). These terroir units can, in turn, be used as input for labeling samples to be analyzed by traditional classification methods. In this work, we describe the methods and the algorithms that permit to produce GIS maps from NMR profiles and apply the described method to the analysis of the geographical distribution of olive oils in an Italian region. In particular, we analyzed by H NMR up to 98 georeferenced olive oil samples produced in the Abruzzo Italian region. By using the first principal component of the NMR variables selected according to the Moran test, we produced a GIS map, in which we identified two regions incidentally corresponding to the provinces of Teramo and Pescara. We then labeled the samples according to the province of provenience and built an LDA model that provides a classification ability up to 99% . A comparison between the variables selected in the geostatistics and classification steps is finally performed. Copyright © 2016 John Wiley& Sons, Ltd.