Electric vehicles have been proliferating in large cities across the world, and we increasingly face challenges in estimating charging demand and in planning EV infrastructure. Focusing on Beijing as a case study, this research uses a novel data-driven method to measure the Charging Demand Indicators (CDI) derived from location-based service big data. Analyses through kernel density function reveal dynamic relations between the spatial patterns of CDI and the distribution of Public Charging Stations (PCS). Spatial match examination is conducted to discover areas of mismatch between charging demand and infrastructure supply. The results expose a CDI pattern which, although largely complies with the city's centripetal structure, demonstrates variations between weekdays and weekends and by EV travel distances. A spatial regression model confirms the influence of urban structure and distribution of amenities on EV charging behavior and suggests that particular land uses and location features have a significant association with EV charging demand. These findings shed light on the understanding of the spatial disparity between the CDI pattern and the current PCS distribution, which could inform future urban policies and planning of EV infrastructure with an emphasis on its coordination with land use, physical layout, and transit.