| Abstract [eng] |
The aim of the master's project is to cluster the countries of the European Union and to assess their competitiveness. Competitiveness assessment is important for identifying disparities between Member States and for seeking the targeted enlargement of the European Union. After the analysis of the scientific literature, the model of the factors determining the competitiveness of the countries was created, consisting of 4 parts - economic, social, country's environment and environmental competitiveness. Using publicly available databases, indicators, which reflected the developed model, were collected. The model of the factors determining the competitiveness of the countries can also be used in other researches. The MissForest algorithm was applied in order to fill in the missing values, which allowed to save valuable data and to use it in further research. The indicators selected according to the model of the factors determining the competitiveness of the countries allowed to cluster countries according to their similarities. In this way, the most competitive and least competitive clusters were identified and differences between clusters were highlighted. In most cases, the clustering results were influenced by the geographical location of the country, although it was not directly included in the study. According to the rankings obtained from the calculated competitiveness index showed that Sweden, Luxembourg, the Netherlands, Denmark and Finland gained the greatest competitive advantage during the period under analysis. The lowest competitive advantage during the analyzed period was in Romania, Bulgaria, Poland, Greece. In order to increase the European Union's as a single territorial unit's global competitiveness, disparities between Member States should be reduced, especially with regard to the lowest-ranking countries. Assuming that competitiveness can be measured by one macroeconomic indicator – gross domestic product per capita, the most important variables that have a significant impact on it have been identified. The machine learning algorithm VSURF was used in order to select the most important variables. 11 indicators were selected, most of them were economic. Comparing the ranks of the European Union countries in the analyzed period, it was noticed that the highest agreement of ranks, measured by the Kendall's coefficient of concordance, was obtained by comparing the ranking of the competitiveness index (using all indicators) with the competitiveness index (using VSURF selected indicators) and this measure was above 0.96. A smaller set of variables could also be used to assess the competitiveness of European Union countries, and gross domestic product has been an important indicator in identifying them. |
