Simon BLAINEY, University of Southampton, United Kingdom
Marcus YOUNG, University of Southampton, United Kingdom
John ARMSTRONG, University of Southampton, United Kingdom
Long-term time series analysis of transport network development has the potential to provide important insights into the factors and processes which influence how transport systems evolve. However, the relevant datasets for such analysis are seldom available in electronic formats which permit easy analysis. This is a particular challenge for data on public transport services, where printed timetables can form the only comprehensive source of information on historic service patterns. This paper compares the relative efficacy of three methods for extracting such data from historic railway timetables: 1) manual geocoding of train routes on an individual basis; 2) crowd-sourced geocoding of train running times in GTFS format; and 3) OCR analysis of scanned timetable data. The challenges and issues associated with using these methods to produce geospatial network-based representations of historic night train services in Europe are discussed. The paper considers both the problems associated with the limitations of the methods, and those associated with the deficiencies and ideosyncracies of the original printed timetables. It concludes by making some suggestions as to the most promising methods for the conversion of other historical timetable data into a format suitable for computational analysis.
Mots clés : Timetable|Railway|Night train|Geocoding|OCR
A102459SB