<![CDATA[Achieving the Sustainable Development Goals (SDGs), including the environmental goals (7, 12, 13, 14, and 15), requires analytical methods that capture long-term national trajectories. Existing studies have not widely used approaches that integrate similarity measurement, clustering, and forecasting. Aims: This study proposes a hybrid framework of similarity measurement, clustering, and forecasting for global SDG trajectories. By comparing cluster structures from historical SDG data with those generated using historical and forecasted trajectories, the study identifies how countries’ development patterns may shift over time. Methodology and results: The framework integrates time-series clustering and predictive modeling. Clustering utilizes both historical data from 2000 to 2025 and a combination of historical and forecasted values, employing DTW variants to measure similarities across 167 countries. K-Means, Agglomerative, and Spectral clustering algorithms are evaluated to identify the most coherent grouping. ARIMA, LSTM, GRU, and Prophet forecasting algorithms are assessed to determine the most accurate SDG score projections for 2026 to 2030. Results show that Soft DTW with K-Means produces the most coherent clusters, and ARIMA yields the lowest forecasting errors. The clustering reveals three groups representing different development pathways: strong SDG index but uneven environmental performance; strong environmental scores despite low SDG index performance; and high SDG performance with moderate environmental outcomes. These patterns highlight diverse sustainability trajectories and the multidimensional nature of global development progress. Conclusion, significance, and impact study: The study validates elastic similarity measures integrated with clustering and forecasting and provides a data-driven decision support framework to improve policy coherence and strengthen international cooperation.]]>
Copyrights © 2026
