A Systematic Approach to Improving Predicted Arrival Time Using Historical Data in Absence of Schedule Reliability
Allen Rawls
April 16, 2008
Committee
Dr. Ron Vetter (chair)
Mr. Albert Eby (Wave Transit)
Abstract
Public transit operations are susceptible to change, both in traffic flow and other conditions that could affect operations such as bridge openings, road floods, and torrential downpours. Traditionally, riders at waiting stops are not informed of the transit vehicles status along the route. Although it is normally not needed for daily transit operations, live location information is particularly useful in cases when vehicles are running behind schedule. This thesis introduces a method for gathering and analyzing historical location and telemetry data of public transit vehicles to better determine estimated arrival time for a vehicle on a closed-loop public transit pattern. The research creates a system for sending real-time locations of transit vehicles to riders through a wide array of mediums including web pages, computer programs, graphical information displays in public locations, mobile phone applications, mobile text messaging, and internet feeds. The system incorporates a weighted estimated arrival time for one route in the city, the University of North Carolina Wilmington campus loop shuttle route, which serves as a working demonstration of these concepts. The approach shows improvement over an arrival time estimate using only average speed
FULL TEXT (pdf)