Predicting the day of storage of dairy products by data combination

Abstract

Abstract Existing methods for tracking changes in the quality of dairy products are characterized by difficulty, time consuming, a large number of calculation procedures and resources, which makes them unsuitable for “on-line” monitoring. In the present work, feature vectors containing color components, spectral indices and physicochemical parameters of the products are used. Methods for selection of informative features based on consistently improving assessments have been applied. Models by principal components and latent variables are derived. It has been proven that the models are adequate and can be used to predict the day of storage of yellow cheese and white brined cheese. The advantage of the proposed data processing procedures, comparing them with those reported in the available literature, is that they require less computational resources, which makes them suitable for use in “on-line” monitoring of the condition of dairy products during storage.