巴士旅行时间预测：从 GTFS 数据中学习.pdf

1、Bus Travel Time Predictions:Learning from the GTFS DataMd Ahnaf ZahinDr.Yaw Adu-GyamfiTitan LaboratoryCivil and Environmental EngineeringUniversity of Missouri-ColumbiaBackgroundConference on Innovations in Travel Analysis and PlanningNumber of vehicle miles driven on US highways has increased by 10

2、.1%over the past 10 years,reaching 274.4 billion in January 2022(FHWA 2022)Missouri was the fourth-highest state for average annual mileage driven,with travelers covering 18,521 miles on average(FHWA 2023)St.Louis saw a decrease in public transit use by 8%,with buses accounting up 64%of all trips ta

3、ken on public transportation in 2018(One STL 2020)In the U.S overall,there were 883 million fewer public-transit rides nationally in the third quarter of 2022 than there were in the same quarter in 2019(APTS 2023)Increasing of congested roads,increased greenhouse gas emissions,longer travel times,an

4、d a general degradation in the quality of lifePassengers wait longer at stops,which increases anxiety,fuel consumption,and pollution,stresses the transportation infrastructure,and decreases accessibility and mobilityMotivationConference on Innovations in Travel Analysis and PlanningUse of GTFS data

5、in our research which needs to be collected in a standardized way through Cloud-based APIs and making it easier to integrate with other systems and services.Important features need to be extracted from the data that can potentially help to predict bus travel times accurately.A modeling framework nee

6、ds to be developed that can predict the average bus travel times across multiple routes precisely and can be computationally fast.Identifying a model architecture that can help travel times predict with long-range dependencies.Expected OutcomesConference on Innovations in Travel Analysis and Plannin