Data science methods for comprehensive assessment of regional economic development

  • 1490 Views
  • 30 Downloads

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License

The paper deals with the assessment of the socio-economic development of Ukrainian regions using Data Science methods and multidimensional analysis, including taxonomy, n-dimensional classification, and ensemble decision trees methods. The methodological bases of economic regions devel¬opment by the economic and mathematical modeling methods were investigated. The necessity of improving and further developing estimation models of the regional economic development using business analytics tools and multidimensional scaling methods was investigated.
The ensemble decision trees methods was applied for the classification model of economic development of the Ukrainian regions according to the conceptual base of the research on regional econom¬ic development. It will increase the quality level of administrative decisions making on regional de¬velopment asymmetry equalization. It is determined that in Ukraine, there is a significant imbalance of regions clusters with high and low economic level. Here was investigated the relationship between the two groups of economic development indicators – the indicators of the economic development regional performance and the group of economic potential. The results of the classification model allow identifying the set of indicators that have significant impact on the overall economic development. The developed ensemble model allows carrying out qualitative recognition and prediction of the state probability of economic development. It will improve the quality of decision making pro¬cesses on equalization of regional development asymmetry.
The further research gives the possibility to develop the system of levers directions of regional development imbalance equalization, to determine priority vectors of sustainable development of both the regions and the country.

view full abstract hide full abstract
    • Figure 1. Концептуальна схема моделювання рівня ЕР регіонів
    • Figure 2. Схема роботи алгоритмів бустингу та беггінгу
    • Figure 3. Результати розрахунку рівня ЕР за регіонами
    • Figure 4. Кореляційна матриця індикаторів
    • Figure 5. Схема крос-валідації
    • Figure 6. Графік впливу факторів
    • Figure 7. Графік впливу факторів
    • Table 1. Результати крос-валідації класифікатора на навчальній виборці
    • Table 2. Результати валідації моделі на тестовій вибірці
    • Table 3. Результати основних метрик валідації моделі
    • Conceptualization
      Liubov Chagovets, Svitlana Prokopovych, Viktor Kholod
    • Funding acquisition
      Liubov Chagovets, Svitlana Prokopovych
    • Methodology
      Liubov Chagovets, Svitlana Prokopovych
    • Visualization
      Liubov Chagovets, Viktor Kholod
    • Writing – original draft
      Liubov Chagovets, Svitlana Prokopovych
    • Writing – review & editing
      Liubov Chagovets, Svitlana Prokopovych
    • Project administration
      Svitlana Prokopovych
    • Supervision
      Svitlana Prokopovych
    • Data curation
      Viktor Kholod
    • Formal Analysis
      Viktor Kholod
    • Software
      Viktor Kholod