| Nigar Ismayilova, Ibrahim Muradov, Nicat Akhundzade Clustering of Countries Based on Multi-Domain Indicators for Sustainable Development |
|---|
| Abstract. Grouping world countries based on different socio-economic and environmental characteristics not only considers the more reliable categorization but also simplifies the decision-making process to support sustainable development goals. Application of unsupervised learning algorithms for clustering of countries based on various indicators gives more comprehensive and reasonable grouping by eliminating the limitations of previous studies focused on single-domain analysis or restricted to regional analysis. For this purpose, a dataset containing 47 global indicators across 267 countries over a 23-year period (2000-2022) were extracted from World Bank Open Data. Indicators used for countries’ clustering involves data representing land use, population trends, CO₂ emissions and energy production/consumption. The first part of the experimental work was focused on data collection and preprocessing, the second phase involved application and comparison of several clustering methods for grouping of the countries. The ability to obtain intra-cluster and inter-cluster comparisons based on global development indicators using clusters derived from experiments offers a great advantage over policymaking and research by identification of transferable strategies for growth, sustainability and reform. |
| Keywords: unsupervised learning clustering, World Bank indicators, Global development patterns, Multi-domain analysis |
 Download PDF |
| DOI: |
