Abstract

The Sustainable Development Goals (SDGs), adopted by the United Nations in 2015, provide a comprehensive framework to address critical global challenges such as poverty, inequality, and environmental sustainability, with the aim of promoting well-being and prosperity for all by 2030. Predicting progress towards SDGs is essential for enabling proactive policy-making and efficient resource allocation, ensuring that interventions are effectively targeted to areas of greatest need. This paper proposes a two-step process for constructing machine learning models to predict SDG indicators. The first step involves a shape-based clustering method that groups countries with similar underlying characteristics, creating more homogeneous clusters for analysis. In the second step, separate machine learning models, including xgboost and LSTM, are built for each cluster, tailored to the specific characteristics of the countries within each of these groups. We applied this workflow to indicator 9.2.1, which measures manufacturing value added as a proportion of GDP. Our comparative study demonstrates that the machine learning models developed using this two-step process significantly outperform the classical ARIMA and exponential smoothing model. Hence the results highlight the potential of this approach to enhance the accuracy of SDG predictions, thereby supporting more effective, data-driven decision-making for sustainable development.