Improved methods to deduct trip legs and mode from travel surveys using wearable GPS devices: A case study from the Greater Copenhagen area

• Published in: Computers, environment and urban systems Vol. 54; pp. 301 - 313
• Main Authors: Rasmussen, Thomas Kjær, Ingvardson, Jesper Bláfoss, Halldórsdóttir, Katrín, Nielsen, Otto Anker
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01-11-2015
• Subjects: GIS; GPS data processing; Handheld GPS; Multi-modal travel survey; Revealed preference data; GIS; Handheld GPS; Revealed preference data; GPS data processing; Multi-modal travel survey
• ISSN: 0198-9715; 1873-7587
• DOI: 10.1016/j.compenvurbsys.2015.04.001

•We develop and test a method to process raw individual-based GPS data.•The method relies on combined fuzzy logic, GIS analyses and feedback algorithms.•High fit rates in the detection of trips, trip legs and mode for Copenhagen case.•92% success rate across five modes with particular good rates for bus and rail.•The results were not highly sensitive to changes in the specification of the input.
GPS data collection has become an important means of investigating travel behaviour. This is because such data ideally provide far more detailed information on route choice and travel patterns over a longer time period than possible from traditional travel survey methods. Wearing a GPS unit is furthermore less requiring for the respondents than filling out (large) questionnaires. It places however high requirements to the post-processing of the data. This study developed and tested a combined fuzzy logic and GIS-based algorithm to process raw GPS data. The algorithm is applied to GPS data collected in the highly complex large-scale multi-modal transport network of the Greater Copenhagen area. It detects trips, trip legs and distinguishes between five modes of transport. The algorithm was validated by comparing with a control questionnaire collected among the same persons and a sensitivity analysis was performed. This showed that the algorithm (i) identified corresponding trip legs for 82% of the reported trip legs, (ii) avoided classifying non-trips such as scatter around activities as trip legs, (iii) identified the correct mode of transport for more than 90% of trip legs, and (iv) were robust towards the specification of the model parameters and thresholds. The method thus makes it possible to use GPS for travel surveys in large-scale multi-modal networks.