MULTICOLLINЕARITY IN REGRESSION ANALYSIS OF WHEAT GLUTEN INDICATOR DURINGITS STORAGE

Abstract

When constructing regression models describing the storage processes of agricultural products, researchers often ignore the effect of multicollinearity factors (the multicollinearity factor effect). In this case, the interpretations of the estimates of the regression model parameters are not adequate. It is suggested to perform multicollinearity diagnostics as one of the steps of the regression analysis. In the case of correlation of factors, it is proposed to construct a regression model by regularizing the LASSO parameters. In this paper a regression model is built, that describes the effect on the quality indicators of wheat grain (gluten) of storage parameters in the granary during cooling (“Average grain temperature in the granary, ºC”, “Humidity of the grain, %”, “Air temperature in the granary, ºC”, “Refrigerant temperature, ºC” and “Volume air flow, m3·h-1”). The parameters of the constructed model are analyzed. The conditions for the long-term storage of cereals are the synthesis of active ventilation by the convective stream of the coolant, turning of loose mass. The developed diagrams of granaries give an opportunity to realize the complex technological processing of products at the expense of efficient distribution of the cooled air flow passing through the raw materials, regulating its gas composition in intergranular spaces, humidity, temperature and intensity of physiological and microbiological processes in these layers.