ABSTRACT
The absence of a robust impact assessment framework has constrained efforts to evaluate how transport infrastructure projects contribute to the Sustainable Development Goals (SDGs). Digital technologies now offer new opportunities to improve how project outcomes are measured and integrated into planning processes. This study proposes a machine learning (ML)–based framework to assess the alignment of transport infrastructure initiatives with SDG-related criteria. A mixed-methods approach was applied, combining benchmarking analysis, expert judgment using the Analytic Hierarchy Process (AHP), and supervised ML modeling. The framework integrates 15 indicators grouped under four SDG pillars of social, economic, environmental, and institutional, weighted through AHP and classified using three ML algorithms: Naïve Bayes, Decision Tree, and Deep Learning/Artificial Neural Network (DL/ANN). Among them, the DL/ANN model achieved the highest accuracy at 86.7%, which demonstrates strong potential for sustainability-based project evaluation. The study contributes academically by bridging SDG indicator systems and ML analytics into a unified assessment framework that connects qualitative judgment with predictive data modeling. Its practical contribution lies in the deployment of an interactive dashboard and simulator that policymakers and planners can use to screen, compare, and prioritize transport projects according to their expected SDG impact. By providing a data-driven and replicable evaluation process, this framework supports evidence-based infrastructure planning and long-term sustainability alignment.
