Utilizing Crowdsourced Data to Extract System-Wide Patterns of Rail Transit Delays

Abstract

Utilization of public transportation depends on reliability and safety of the transit system, which relate to how the delays occur and propagate on the transit network. Similar to roadway transportation, the delays at rail transit systems can also be categorized as non-recurrent and recurrent. This study focuses on the latter in order to reveal delay patterns and knock-on effects occurring on the transit network. For this purpose, we utilize crowdsourced data for the analysis of the recurrent delays occurring at Long Island Rail Road (LIRR) commuter rail system in the State of New York. We conduct an exploratory analysis of the real time transit information app onTime data and calculate delay estimates for different origin-destination (OD) rail service lines and individual stations. The data consists of real-time estimated delays to the stops, based on the difference between the scheduled arrival time and the actual position of the app user. The results indicate that there are distinct differences between delay patterns based on stations proximity to bottleneck stations and the priorities of individual OD lines on the overall transit network. Moreover, findings imply that transportation authority might be adjusting right of way of routes systematically at bottleneck locations to balance and compensate with the accumulated delays.