A. A. Dubovitskiy, E. A. Klimentova, V. A. Shatskiy
Michurinsk State Agrarian University, Michurinsk, Tambov region, Russia
E-mail:
Volume 26 No. 10
Date of paper submission: 20.07.2026, date of review: 28.08.2026, date of acceptance: 16.09.2026.
Published: 10/31/2026
Abstract. The purpose of the study is to assess spatial heterogeneity and price differentiation factors in the milk market. Methods. The research is based on the use of elements of statistical and economic analysis in a combination of empirical, descriptive, analytical, dynamic, comparative spatial and predictive approaches. The identification of factors that have a decisive impact on regional price differentiation in the milk market is based on neural network analysis using a multilayer perceptron model, which allows for the study of hidden representations through nonlinear transformations. The scientific novelty. The parameters of spatial heterogeneity are determined and the factors that have a decisive influence on the regional differentiation of prices in the milk market are identified, recommendations for improving the placement of dairy cattle breeding at the regional level are formulated. Results. In the course of the study, key trends in the development of dairy cattle breeding were identified, characterized by an increase in milk production and a simultaneous reduction in livestock; parameters of changes in milk prices were determined, characterized by positive dynamics, especially during the second half of 2024 and early 2025, with the formation of the “overproduction paradox” in early 2026, when production growth exceeded effective demand, which led to lower prices. The neural network analysis made it possible to localize the factors that have the most significant impact on the formation of the forecast of prices for the regional raw milk market, including: the proportion of the urban population (NI = 0.079), average per capita monetary income (NI = 0.074), the proportion of the population younger than working age (NI = 0.073), milk yield per cow (NI = 0.071), the population of the region (NI = 0.067) and the gross milk yield in the region (NI = 0.068). The importance of these factors as a basis for forecasting prices for raw milk sold by agricultural organizations and predictors of improving market conditions is substantiated.
Keywords: agro-food system, region, dairy cattle breeding, location, milk market, profitability, price, forecasting, machine learning, neural networks
Acknowledgements. The research was carried out at the expense of the grant of the Russian Science Foundation No. 25-28-01313 (https://rscf.ru/project/25-28-01313/).
For citation: Dubovitskiy A. A., Klimentova E. A., Shatskiy V. A. Spatial estimation of the heterogeneity of prices in the milk market using the multilayer perceptron model. Agrarian Bulletin of the Urals. 2026; 26 (10): 2056‒2069. https://doi.org/10.32417/1997-4868-2026-26-10-2056-2069 (In Russ.)
Download the full text of the article









