Fourier Transform Infrared Spectroscopy Combined with Chemometrics in Goat Milk Authentication

Chemometrics improves the efficiency of the Fourier transform infrared (FTIR) spectroscopy as a method for authentication of pasteurized goat's milk in its mixes with cow's milk. The research featured mixes of two brands of goat's and cow's milk in various proportions, from 1 to 90%. The chemometric modeling involved the method of partial least squares regression (PLS) to predict the authenticity percentage based on preprocessed spectral data. A five-fold cross-check was performed to verify the PLS model. The dataset was divided into five groups, each of which was used to train and test the model. The prediction error was quantified using the standard deviation (RMSE), which measured the mean deviation of the predicted values from the actual ones. The RMSE was calculated for value, and the mean RMSE for all values served as an indicator of prediction accuracy. The method of bar graphs visualized the RMSE for each fold; the cumulative variance was explained by the PLS components. The PLS model obtained by the FTIR spectroscopy demonstrated that the mean error remained below 3–5% even in the worst scenario. The IR-Fourier spectroscopy proved to be a fast, cheap, and non-destructive authentication tool for goat's milk.