Classification of Raw Milks Using Pattern Recognition Methods

The objective of this study was to assess the relationships between physicochemical and
microbiological properties of raw milk and the use of multivariate statistical analysis such as
principal component analysis (PCA), hierarchical cluster analysis (HCA) and stepwise discriminant analysis (SDA) for pattern recognition and classification it. In this study, 48 raw milk samples were collected from some dairy herds of Mashhad. Samples were analyzed for the microbiological and physicochemical properties. PCA, HCA and SDA were applied to estimate the usefulness of the physicochemical and microbiological parameters for the differentiation and classification of raw milk using. The results of PCA shown the seven principal components explained 93.65% of total system variance. The PCA method permits a good classification between raw milk samples on the basis of the first three PCs. HCA classified physicochemical and microbiological properties of raw milk into three main groups that confirmed the correlation between the studied variables obtained by PCA. Using SDA it was determined which variables best classified the raw milk samples according to their quality. Finally, the classification functions allowed the correct classification of 91.7% of the raw milk samples. Due to the direct effect of raw milk quality on dairy products quality and consumer health, the quality of raw milk has special importance in the dairy industry. Therefore, classification of raw milk based on the quality characteristics will help to determine the price of raw milk and to produce high quality dairy products.