Uncovering spatiotemporal patterns and influencing factors of electric vehicle battery capacity preferences at the city-level in China

In the context of rapid electric vehicle (EV) expansion, single-vehicle battery capacity has emerged as a core variable linking total battery demand, life-cycle carbon emissions, and critical mineral security. However, a systematic empirical understanding of the statistical characteristics and evolutionary mechanisms of battery capacity at high granularity levels remains limited. Leveraging a multi-dimensional, city-level dataset that encompasses EV sales and characteristics, climatic conditions, infrastructure, socioeconomic factors, and market conditions from 2019 to 2024, we developed a high-resolution, spatiotemporally explicit, interpretable machine learning framework that combines random forest models with Shapley additive explanations analysis to identify key drivers of battery capacity preferences. The results indicate that, at the national level, the average battery capacity of battery electric vehicles in China increased from 46.9 kWh in 2019 to 55.9 kWh in 2024 during the study period, yet it exhibited significant spatiotemporal divergence and regional effects across different city clusters. Vehicle segment preference, market maturity, and climatic conditions are the decisive factors shaping battery capacity preferences, with the impacts of these variables showing substantial spatial heterogeneity. Public charging infrastructure in most cities influences capacity selection not by directly alleviating range anxiety, but by cultivating a market environment conducive to EV penetration. These findings offer data-driven support for managing EV battery demand and related impacts towards sustainable electrification.