Promoting urban energy transitions: Lessons from interpretable machine learning with evidence from China

Promoting urban energy transitions is essential for global governments to address climate change and achieve low-carbon goals. To identify key drivers and optimal strategies for accelerating urban energy transitions, this study analyzes data from Chinese cities (2003–2019) using machine learning and interpretable algorithms to examine transition pathways from the perspectives of energy system performance (ESP) and transition readiness (TR). A Monte Carlo approach with kernel density estimation is further applied to simulate future transition scenarios from both dimensions. The results indicate that most drivers of the energy transition in cities leverage the synergistic effects of both ESP and TR, while a few only advance energy transitions through one or the other. Simulations suggest that under current development paths, the pace of China's urban energy transition will gradually slow down, and further analysis reveals substantial regional, resource endowment, and urban functional heterogeneity in ESP and TR. The evidence from Chinese cities shows that cities face choices between prioritizing ESP or TR in their energy transitions. The complex scenario simulations reveal that Chinese cities retain strong potential to advance the urban energy transition. However, realizing this potential requires cities to make greater efforts tailored to their specific conditions in promoting economic development, managing population distribution, and strengthening technological innovation. Governments should consider the heterogeneous characteristics and unique realities of cities, refraining from applying a one-size-fits-all approach to successfully meet urban energy transition goals.