Novel AVL-APC data analytics platform offers Pittsburgh transit system a systematic way for users and decision makers to understand system performance from many aspects of service quality, including passenger waiting time, stop-skipping frequency, bus bunching level, bus travel time, on-time performance, and bus fullness.
Understanding Transit System Performance Using AVL-APC Data: An Analytics Platform with Case Studies for the Pittsburgh Region
The platform offers a systematic way for users and decision makers to understand system performance from many aspects of service quality, including passenger waiting time, stop-skipping frequency, bus bunching level, bus travel time, on-time performance, and bus fullness. By combining the advantages of web applications (such as low cost for development and maintenance, convenience for access anywhere and anytime, customizability, and scalability) and of AVL-APC data (such as low cost, broad coverage, and high quality) the data analytics platform helps automate the public transit data requests and analysis process with high spatio-temporal resolutions. There are two major advantages when compared to similar platforms:
- All data fetching, visualization, and calculations in this platform use archived AVL-APC data. The dataset is able to provide historical bus arrival, departure, and load data accurately down to the trip and stop level, so queries may be performed for a single trip, stop, or route, up to the transit system in the Pittsburgh city area as a whole.
- The platform offers a systematic way for users and decision makers to examine system performance from many aspects of service quality, such as passenger waiting time, bunching, bus trips that skip stops due to heavy loads (stop-skipping), crowding level, travel time, and on-time performance. Those metrics can be accurately estimated in the platform at any desired spatiotemporal resolution, allowing a better understanding of system performance.
The AVL-APC data from September 2012 to March 2016 were archived in a database to support the development of a user-friendly web application that allows both users and managers to interactively query bus performance metrics for any bus routes, stops, or trips for any time period.
The study investigated two case studies for using the data analytics platform: Bus Bunching and Schedule Change Analysis.
Case 1: Bus Bunching in Pittsburgh
This analysis compares the incidence of bunching between two high-ridership routes operated by the Port Authority of Allegheny County (PAAC) in Pittsburgh with different operational characteristics. Route P1 is a bus rapid transit (BRT) route that operates on a dedicated busway. Route 61C is a "key corridor" route that operates in mixed traffic connecting several of Pittsburgh’s business districts, including the central business district (CBD) and Oakland. Using AVL-APC data available through the analytics platform, hourly heat maps of bunching incidence for weekdays in the period of March 1 – March 31, 2016, were created for both Route 61C outbound and Route P1 outbound for the period from 12:00 to 10:00 p.m.
Case 2: Schedule Change Analysis for Route 61B
A comprehensive analysis was performed on Route 61B inbound, which travels from Braddock through Oakland to downtown Pittsburgh. Passenger waiting time, bunching level, crowding level, schedule deviation, and stop-skipping under two different bus schedules were studied and compared. The AVL-APC data used before the schedule change cover November 9–20, 2015 (except weekends), and the data after the schedule change cover November 30–December 11, 2015 (except weekends).
- Case 1: It was found that the incidence of bus bunching is heavily impacted by the location on the route as well as the time of day, and the bunching problem is more severe for bus routes operating in mixed traffic than for bus rapid transit, which operates along a dedicated busway.
- Case 2: The results showed that the new schedule leads to approximately 20 percent less average absolute schedule deviation, and approximately 10 percent less average passenger waiting time.