Spatiotemporal Analysis of Electric Taxi Service Mode Based on Gaussian Mixture Model

Vehicle electrification is important for promoting green transportation. Using a large-scale Electric Taxi (ET) operation data set in Shenzhen, the study explored the spatiotemporal distribution of ETs’ service mode considering different shift modes. Specifically, segmenting the time sequence of ET orders into different work slices (WSs), Gaussian mixture models were used to cluster these WSs of single-shift and double-shift ETs into 9 and 12 spatiotemporal feature-based clusters, respectively. Subsequent analysis was conducted on the service modes combined by different clusters for both shifts. Results show the following: (1) ETs operated in central core areas often receive higher fares. (2) Implementing service mode combining 2 or 3 clusters could help increase the fare for ET drivers. (3) Double-shift drivers in core areas may experience fare imbalance and potential charging peak at shift hours. The outcome offers insights for customizing ET operation strategies, such as subsidy design and order matching algorithms.