Content of collections

Methods for adjusting predictive decisions

Kulyk М.M., Dr. Sci. (Engin.), Academician of the National Academy of Sciences of Ukraine
Institute of General Energy of the National Academy of Sciences of Ukraine, 172 Antonovycha St., Kyiv, 03680, Ukraine
Language: Ukrainian
Source: The Problems of General Energy, 2014, 2(37):5-12
Section: Systemic studies and complex problems of the energy sector
UDC: 620.9
Received: 24.04.2014
Published: 29.04.2014

Abstract:

We design a new mathematical model and computational methods for predicting various indicators characterizing the development of economy and society, such as gross domestic product, the consumption of all kinds of fuels and power resources, desired investments, greenhouse gas emissions, population size, and other indicators. This model represents an overdetermined system of algebraic equations derived for making predictions at the macrolevel (T-level) and sectorial level (D-level). The initial system of algebraic equations is transformed by multiplying the system matrix by the corresponding transposed matrix. For the resulting system of any order, the analytical solutions are obtained. It is determined that the solutions obtained do not provide the coincidence (consistency) of prediction at the T-level and the sum of predictions at the D-level. A special iterative procedure is developed, and resulting analytical solutions, which comply with the consistency requirements, are obtained for systems of any order. The multivaluedness of the solutions obtained is disclosed, and the best predictions, minimizing the sum of squared errors on the set of possible predictive decisions, are found. On the basis of the proposed mathematical model, two methods for adjusting predictions at the T-level and D-level are developed. The comparative analysis of these methods is performed, and recommendations for the preferable use of these methods are provided. We also give an example of adjusting predictions of the demand for electric power in Ukraine up to 2030.

Keywords: prediction, indicator, mathematical model, rectangular matrix, iterative process, adjustment, demand.

Downloads:

Full text (PDF)