The 20th and 21st joint COTA International Conference of Transportation Professionals (CICTP 2020-21)

Derivation of Rail Passenger Traveling Information (RT) acts an important role in rail-service improving. In early studies, researchers usually derived RT (including entrance-station, exit-station, and transfer-station) with Automatic Fare Collection System Smart Card Data (AFCD), but the timeliness and integrity of trace cannot be guaranteed. The signaling data break the limits of timeliness and integrity of trace, while traditional 2/3G signaling data is also hard to support complete identification of RT for its long trigger-interval. The latest 4G signaling data provides a chance to more accurate RT-derivation with a shorter trigger-interval. So, rooting in 4G signaling data, this paper proposes a method of calculating station-stickiness and building an SVM classifier of preparing data. Then, derive RT with a rule of judging entrance-station or exit-station assisting, finally carry a detailed validation. When validating our result with travel-logs of 21 volunteers in 7 consecutive days, we find that the average difference of entrance-station in 4 rail-stations reach 8.40%, and the average difference of exit-station in 4 rail-stations only 4.47%, which demonstrates the plausibility and efficiency to accurately assess of RT with our method. Comparing with former methods and data sources, our method and data source provide a more efficient and accurate RT-derivation.

CICTP