A Methodology for Denoising and Generating Bus Infrastructure Data

Abstract

Together with the availability of new mobility data, the development of new intelligent transport systems (ITS) have increased, in order to provide new key performance indicators toward the improvement of the management of traffic awareness in cities. ITS rely on accurate transit infrastructure data that often contain erroneous information (e.g., inconsistencies or is out of date). In this paper, we propose a new methodology that makes use of GPS traces to automatically detect or correct bus stop locations, reconstruct bus route shapes, and estimate time schedules. The methodology performs different steps: 1) data cleaning and detection of trips; 2) bus stop extraction through data mining techniques; 3) route shape reconstruction; and 4) time schedule estimation. A case study using real GPS data from the City of Dublin, Ireland, is performed.