Neural Network and Regression Analysis of the Dependence of the Ranking Score of Organoleptic Characteristics on the Food System Composition

Abstract

Aim: The aim of this research is to develop the technology for producing antianemic smoothie for pregnant women based on the optimization of the composition and organoleptic characteristics of the final product. Methods and Materials: The relationship of organoleptic criterion with the quantitative composition of the formulation is identified by neural network and regression analysis of the ranking score of organoleptic characteristics. The model parameters are obtained by means of “Statistica†software package. Convolution of balancing index and sensory evaluation is proposed in the form of a multiplicative function of desirability. The weight optimizing problem of the smoothie composition was solved by means of MathCAD scripts. Results and Discussion: At that, wheatgrass juice is an alternative source of iron supplement in the recipe of the smoothie. When modeling the weight composition of the developed product, particular attention was paid to the physiological need of pregnant women in nutrients and their content in food sources such as kiwi, grapes, yogurt, wheat germ juice, and honey. Mathematical optimization model of the smoothie composition for pregnant women needs to take into account the organoleptic characteristics of the combined product. The sensitivity of the network was the main criterion for the selection of ingredient composition that maximally contributes to sensory evaluation when mixing a smoothie. Conclusion: We came up to the conclusion that the most important components of the smoothie mix (in the context of organoleptic characteristics) are wheatgrass juice and yoghurt. In consequence of the research, we have developed the methodology of mathematical modeling of the smoothie composition for pregnant women, which meets most fully the following requirements: Optimal content of feedstuff with antianemic properties, and high organoleptic characteristics of the final product.