Assessing the Level of Production Potential of the Regions of Ukraine with Use of Neural Networks

Abstract

The problem of improvement of methods of estimation of level of development of regions of Ukraine on production potential on the basis of building of integrated indicators and carrying out of clustering has been considered. It has been proposed to apply an integrated approach to analysis of the aggregate of indicators of the regions’ performance, characterizing the production potential, with the purpose of constructing integrated indicators by different approaches and with subsequent clustering of regions using the Kohonen neural networks. Use of the Kohonen maps along with database clustering allowed to design multidimensional data in a two-dimensional space, to carry out an analysis of the resulting cluster system, and to improve the results of clustering by selecting the optimal quantity of split groups. The convenient form of visualization of results of clustering provides for localizing features and making the corresponding corrections in the rating list, proceeding from expert judgments.

Keywords

• production potential
• Kohonen neural network
• level of development of regions
• integral estimation