Пропустить команды ленты
Пропустить до основного контента
SharePoint
Перейти вверх

HYBRID INTELLIGENT SYSTEM OF FORECASTING OF THE SOCIO-ECONOMIC DEVELOPMENT OF THE COUNTRY

Наименование публикации:HYBRID INTELLIGENT SYSTEM OF FORECASTING OF THE SOCIO-ECONOMIC DEVELOPMENT OF THE COUNTRYАвторы:Данько Т. П. , Китова О. В., Колмаков И. Б., Дьяконова Л. П., Гришина О. А.
Секерин В. Д.
Тематическая область:Экономика и экономические науки
Вид публикации:Статья в журнале
Электронная публикация:НетЯзык издания:РусскийГод издания:2016Страна издания: Германия Наименование журнала или сборника:IJABERНомер журнала (с указанием года):9 (2016)Код ISSN или ISBN:5755-5766Количество страниц:12Количество печатных листов:0,8Индексация:ScopusАннотация (реферат):

Implementation of the models of short- and medium-term forecasting of macroeconomic indicators for the detection of trends in the development of a whole country or a particular region is an urgent and important task in the field of macroeconomic research.

The article presents the author's concept of development of scenario variant predictions based on simulation regression and factor models complemented by the author's hybrid approach to forecasting of the indicators of socio-economic development with the use of neural networks. The hybrid intellectual and economic system was developed to support the federal and regional levels of the Russian economy with means of analysis and forecasting, which implements the principles of management and decision-making on the choice of a model based on the verification of forecast calculations. The system carried out large-scale computer tests on variative forecasting of the complex of more than 600 indicators of socio-economic development of Russia based on the scenario indicator values given by experts, such as the key interest rate, international oil prices, the money supply growth rate, the change in foreign exchange reserves. After identifying a set of scenario indicators, the projected trajectory is calculates. Comparison of the reference trajectory with the perturbed trajectory (by the known scenario conditions) allows experts to obtain quantitative estimates of the differences and give their assessment. Basic forecast calculations are carried out in the framework of the econometric system and give satisfactory values for the quality and accuracy of 66-70% indicators. Neural network models are built for other indicators, which allow to improve the quality and accuracy for another 10% of the total number of indicators. The development of ​the system is anticipated in the area of improving the neural network models and connectivity ​of the forecast modules implementing decision fork algorithms.

Перейти к списку публикаций