Exploration et traitement multidonnées appliqués à des corridors d'autobus

Passive information collection and their low costs, considering the large amount of data generated, explain the increasingly more widespread implementation of automatic data collection systems. These data represent a great potential to improve the characterisation of supply and demand of public transit for operational planning. However, in addition to their treatment, the integration of these data from multiple systems is a major challenge. This study develops tools and systematic methods in order to exploit these types of data. The completion of the study, in an exploratory research and analysis context, has the perspective of integration of multiple-source data. This project uses the crowded Saint-Michel and Pie-IX corridors of Montreal's bus network as examples for the application of the presented methods. First, the planned network is estimated with GTFS data, used for traveller information: the passage-stop data are compressed by sequences of blocks and runs. Non-commercial runs and line-patterns are inferred from GTFS data. By classifying the network's lines, the evaluation of buses during time reveals the resources needed for the service. Next, arrival and departure time and passenger movements can be associated to stops with GPS tracks and passenger counting data (AVL-APC).