Comparing GPS data quality and Travel Survey data

Comparing GPS data quality and Travel
Survey data quality:
A case study in Copenhagen
Henrik Harder*, Kristian Hegner Reinau, Noam Shoval, Amit Birenboim, Christian
Hansen Overgård, Otto Anker Nielsen, Thomas Kjær Rasmussen, Jesper Bláfoss
Ingvardson.
*Corresponding author:
Department for Architecture, Design and Media Technology
Aalborg University
Abstract for Trafikdage 2015
GPS tracking is on the way from being a method, which is used to support travel diaries, to
becoming a method, which has the potential of replacing travel diaries. As argued by (Reinau,
Harder, & Overgaard, 2014) , there is currently little agreement in the literature on how to
design a GPS tracking study. (Reinau, Harder, & Overgaard, 2014) ‘s V-model for designing a
GPS tracking data survey constituted a first step towards a best practice, but there is still a
need for a methods to specifically evaluate the quality of GPS data, and also for a method to
compare the results of GPS tracking surveys to travel diary surveys.
This paper therefore contributes two things to the literature; first it develops a number of
indexes that can be used to evaluate the quality of GPS data. These include the number of valid
activity hours, the identification of random outliers and speed outliers, and temporal patterns,
as well examination of influence of respondent characteristics, such as variation in data
quality between different age groups and based on the family roles of participants. Second,
the paper develops a number of indexes to compare the quality of GPS data to the quality of
travel survey data. These includes how many trips each method capture, and also the success
rate seen in relation to specific modes of transport, i.e. are one of the methods better for
capturing certain modes of transport?
The potential of these indexes are explored and illustrated on the results from a case study,
where two GPS tracking data collections and a travel survey were done, and it is explored how
different GPS tracking setups influence the quality of the collected GPS data, as well as how
the collected GPS data compares to the travel survey data. The case study is the ACTUM data
collection done in Copenhagen as part of the ACTUM project, in which approximately 100
households in Copenhagen was tracked, two different tracking setups were deployed, each on
approximately 50 households, and all 100 households had to fill out a web-based travel
survey also.
References
Reinau, K. H., Harder, H., & Overgaard, C. H. (2014). Horses for courses: Designing a GPS
tracking data collection. In H. J. P. Timmermans, & S. Rasouli (Eds.), Mobile technologies
for activity-travel data collection and analysis (pp. 36-52) IGI Global.