STATISTICAL ANALYSIS OF AGRICULTURAL EFFICIENCY USING MACHINE LEARNING METHODS

DOI 10.32651/229-100

Issue № 9, 2022, article № 15, pages 100-105

Section: Problems of agroeconomic researches

Language: Russian

Original language title: СТАТИСТИЧЕСКИЙ АНАЛИЗ ЭФФЕКТИВНОСТИ СЕЛЬСКОГО ХОЗЯЙСТВА С ПРИМЕНЕНИЕМ МЕТОДОВ МАШИННОГО ОБУЧЕНИЯ

Keywords: AGRICULTURE, EFFICIENCY, PRODUCTIVITY, AUTO REGRESSION, REGRESSION, CLUSTER ANALYSIS

Abstract: In the context of the ever-increasing sanctions pressure on Russia, it is necessary to have tools for forecasting key indicators of agricultural development, determining the degree of differentiation of regions by the level of efficiency and statistically significant models of the regression relationship of factor and performance characteristics of the industry. These tools and methods of statistical analysis will allow not only to support the solution of issues of food security and export potential, but also the development of rural areas of the Russian Federation as a whole and its individual subjects. This study confirms the presence of cyclical fluctuations in grain yields in Russia every 10 years. An accurate forecast of grain yields for the next two years, as one of the most important export products of our country, has been built. The auto regression model proposed in the article can be used to predict performance indicators not only at the country level, but also at the level of a region, enterprise or individual unit. Based on the multiple regression model, the key factors determining the change in grain yield are identified. High indicators of association between dependent and independent variables (correlation and determination coefficients) allowed us to establish that among the selected factors, the most significant impact on the yield during the study period from 2006 to 2020 is provided by the introduction of mineral fertilizers, the amount of state support, the ratio of investments of regional and federal budgets, climatic conditions. As a result of the implementation of cluster analysis, 5 clusters were identified according to the level of agricultural efficiency. Out of the total of 77 regions studied, 42 regions were assigned to clusters with a low level of agricultural efficiency. These subjects are of particular interest because they have a high potential to increase efficiency, and hence the volume of agricultural production.

Authors: Demichev Vadim Vladimirovich