AI-Driven Operational Sustainable Governance Framework for Municipal Performance Optimization: Empirical Evidence from Jordan

Improving municipal efficiency and governance resilience has become increasingly critical in the face of fiscal pressure, demographic growth, service delivery complexity, and the growing demand for evidence-based operational sustainability and municipal efficiency. However, prior studies have largely examined municipal efficiency evaluation, predictive analytics, and optimization techniques separately, limiting their practical value for integrated decision support in public sector governance. To address this gap, this study proposes an integrated analytical framework combining data envelopment analysis (DEA), machine learning, and metaheuristic optimization to support operational sustainability and governance resilience in Jordan. The empirical analysis is based on panel data from 20 Jordanian municipalities covering the period 2011–2020. First, input-oriented CCR, BCC, and slack-based measure DEA models were employed to estimate technical, pure technical, scale, and slack-adjusted efficiency. Second, random forest, XGBoost, support vector machine, and artificial neural network models were employed to predict the scores for efficiency derived from DEA using validation procedures based on time-based criteria. Third, a particle swarm optimization (PSO) technique was employed to determine operationally and financially viable solutions for allocating municipal resources. The findings reveal significant variation in terms of efficiency among the municipalities. Furthermore, numerous possibilities exist to improve the sustainability and performance of municipal governance. In terms of prediction models, random forest exhibited higher predictive accuracy compared to other models.