STIETR: Spatial-temporal Intelligent E-Taxi Recommendation System Using GPS Trajectories

E-taxies (ETs) are facing great challenges such as short driving range, long charging time and sparse charging stations, thus hamper its acceptance by fuel taxi drivers. This study presents a novel spatial-temporal intelligent recommendation system for e-taxi drivers to improve their net revenue. The knowledge of taxi travels, including the probability of picking-up passengers and destinations, is learned from fuel taxies' raw GPS trajectories to estimate the expected net revenue (ENR) of the e-taxi. Consecutive actions of ET drivers are modeled by action trees to find the best route going to a recharge or cruising along some roads. An online recommendation querying subsystem is developed for high-efficient real-time recommendation. An experiment in Shenzhen using GPS trajectories of 16, 146 fuel taxies is conducted to evaluate the performance. The result shows that, by adopting the proposed system, the net revenue per unit working time of the ET drivers is up to 91.4% better than real-world fuel taxi drivers.