Forecast of development of agrarian sector of economy with the use of artificial neural nets

Abstract

The purpose. To substantiate the theoretical and methodical approach of system forecasting of agricultural production on the basis of the application of artificial neural networks, taking into account quantitative and qualitative factors of influence on the development of the agrarian sector of the economy. Methods. Economic-statistical, calculation-constructive, monographic, comparative, correlation analysis, artificial neural networks. Results. The theoretical and methodical aspects of using the artificial neural network method in the problems of forecasting the development of the agrarian sector of the economy for achieving greater objectivity and accuracy are considered. The systematic analysis of the main production and economic indicators of the agrarian sector of Ukraine for 2000-2017 was carried out. The quantitative and qualitative factors influencing the development of the agrarian sector of the economy were determined. A model for forecasting the development of the agrarian sector of the Ukrainian economy on the basis of an artificial neural network of architectural type multi-layer perceptron is created, which ensures the minimum deviations of the obtained results from the actual values ​​of these indicators. Conclusions. The results of research confirm the possibility of practical use of the model of the ANN developed on the basis of system analysis for adaptive forecasting of the agrarian sector of Ukraine’s economy in the medium and long-term perspective. The proposed model takes into account the factors of change of sown areas, livestock, crop yields, animal productivity and resource provision of agrarian production. A methodical approach is developed for the practical use of artificial neural networks for adaptive forecasting of the agrarian sector of Ukraine’s economy in substantiating different strategies and assessing their impact on the economic, social and ecological state of the industry.