Chemometric techniques for the prediction of milk composition from MIR spectral data: A review

Chemometrics; use of statistical models to characterise and understand complex chemical systems/samples, is an advancing field. In the dairy industry, the accurate prediction of milk composition involves combining mid-infrared spectroscopy with chemometric techniques for the evaluation of major constituents of milk. The increased interest in determination of detailed composition of dairy products, alongside emerging and more-widespread use of chemometric methodologies, have generated continuous improvement in predictive models for this application. Herein the main chemometric techniques employed for the study of milk composition are described, compared and discussed. The capability of emerging technologies to improve predictive accuracy of models, over the gold standard technique Partial Least-Squares Regression, to provide recommendation for future directions in this area are particularly emphasised. This review will be of particular interest to researchers involved in milk and dairy product composition, alongside all working in application of statistical methodologies to complex chemical systems.