Purpose. The creation of the information apparatus for short-term forecasting of key indicators of economic entities (EE) industrial activity based on combinatorial and neural network algorithms of the group method of data handling (GMDH). Methodology. This work presents a methodology for short-term forecasting of main economic indicators of industrial activity of an economic entity using modifications in the classical algorithm of the GMDH, comparing the effectiveness of combinatorial and neural network algorithms of the GMDH. Findings. The analysis of forecasting methods reasoned the choice of the neural network algorithm of the GMDH for the research. The optimal model structure has been compiled and the dependence of the output parameters on the selected input parameters of the system has been derived. The analysis of economic indicators of an agricultural holding has been done. The research results are presented via graphic visualization. The effectiveness of the application of the modified GMDH algorithms for forecasting production trends for up to 3 years has been proved. The regularities of the data changes were revealed. The accuracy of the results was evaluated. Originality. On the basis of the content analysis of the main indicators characterizing the agricultural holding industrial activity, new approaches to the construction of optimal models were proposed, taking into account characteristics of the subject area, which improves the quality of managerial decision making. Information technology for short-term forecasting based on neural network algorithm of the GMDH has been developed. Practical value. The proposed technique allowed us to conduct the short-term forecasting of main economic indicators of an economic entity industrial activity (selling goods and services, cost of the sold products, gross profit, net profit; commercial, general and administrative expenses). The obtained accuracy assessment of the forecasting results is within the range from 1 to 5 %.

  Повний текст PDF - 1.008 Mb    Зміст випуску     Цитування публікації

  Якщо, ви не знайшли інформацію про автора(ів) публікації, маєте бажання виправити або відобразити більш докладну інформацію про науковців України запрошуємо заповнити "Анкету науковця"