@article{4812,
  author = {Jie Wei Zhang},
  title = {SFU-EVP: A Longitudinal Open-Access Dataset of Electric Vehicle Charging Sessions for Grid and Behavioral Analysis},
  journal = {Electronic Devices},
  year = {2026},
  volume = {15},
  number = {2},
  doi = {https://doi.org/10.6025/ed/2026/15/2/72-103},
  url = {https://www.dline.info/ed/fulltext/v15n2/edv15n2_2.pdf},
  abstract = {The rapid adoption of electric vehicles (EVs) has intensified demands on charging infrastructure, yet
conventional performance metrics based on energy delivery alone fail to capture operational inefficiencies
such as post charging idle occupation. This study investigates station level utilization and idle time behavior
in campus EV charging infrastructure using the SFU-EVP longitudinal dataset, comprising 91,518 charging
sessions across 32 stations at Simon Fraser University. Three temporal indicators such ascharging time
utilization, occupancy utilization, and idle time ratio are combined with demand and throughput measures
to construct a seven dimensional station performance feature space. Descriptive analysis reveals that
approximately 23.57% of total plugged in time is idle, with median occupancy utilization (26.64%) exceeding
median charging utilization (20.78%), confirming pervasive post-charging occupation. Principal Component
Analysis captures 86.01% of station level variation in two components, and K-Means clustering (silhouette
score ï‚» 0.50) identifies two distinct station typologies: underutilized/low demand stations (28 stations,
16.96% charging utilization, 24.82% idle ratio) and high performing/high demand stations (4 stations,
46.93% charging utilization, 19.38% idle ratio). Only four stations, 12.5% of the infrastructure, account for
approximately 43.6% of all sessions, demonstrating a strong demand concentration effect accompanied by
greater behavioral efficiency. Hierarchical clustering validates the two group structure. The findings indicate
that effective infrastructure planning must distinguish between capacity expansion at high demand hubs
and behavioral interventions to reduce idle occupancy across the broader network, shifting analytical focus
from how much infrastructure is required to how effectively existing infrastructure is utilized.},
}