Methodological aspects of using geocoded data from mobile devices in transportation research

Abstract

There is an increasing need for information on human mobility patterns in various fields of research from health research to urban planning and civil engineering, and particularly in transport research. Traditionally, research on spatial mobility and travel behaviour has been collected using questionnaires, observations, censuses and so on. Recently, geocoded spatial data from mobile phones and GPS devices have been increasingly utilised to analyse and understand the spatial mobility patterns and travel behaviour of citizens. In this special issue in Journal of Location-Based Services researchers make use of, for example, Call Detail Records (CDR) of mobile operators or Floating Taxi Data (FTD) to identify and map mobility trajectories for mobility studies and transportation research. The data from mobile phones and GPS devices are substantially different from the traditional data sources, which poses new methodological challenges for researchers. Most importantly, the data tends to be secondary, meaning that it has not been collected purposefully for research but is created as a by-product of other activities (Chen et al. 2016; Schwanen 2016). For example, the CDR data are originally generated to keep account of calls and prepare telephone bills, while FTD is collected to organise the logistics of taxi companies. Traditionally transport-related questionnaires and other methods in transport studies primarily involve purposeful questions being asked from a limited number of people about precise activities, trajectories, means of transport, motivations for travelling, etc. These questionnaire-based studies may be planned with a representative sample and adequate questions, which makes them an excellent research material. Compared to these, large secondary data-sets from technical devices generally have a lot of respondents and geolocated data points, but they are often less precise and are/can be only indirectly related to the research objectives (Calabrese et al. 2013; Järv, Ahas, and Witlox 2014). Despite the potential quality issues, the new types of digital data with a large sample size are appealing to researchers because such data are easier to obtain and often cheaper than separate questionnaires. They are also less intrusive and do not require time and effort from volunteers to collect data. Additionally, computational methods needed to process these data are developing at a quick pace and there are a large number of enthusiastic researchers and IT companies that develop business models based on the new data. The improvements in digital data flow makes data collection cycles faster and smoother. In relation to these developments, the Eurostat feasibility study ‘BIG data’ that was composed in 2014 (Eurostat 2014), and several other solid studies, highlight the need for a systematic development of methods regarding new data and validation procedures (Bernardin et al. 2017; Toole et al. 2015). This special issue of the Journal of Location-Based Services has been prepared on the basis of presentations made during the biennial conference Mobile Tartu (http://mobiletartu.ut.ee) that took place from 29 June to 01 July 2016 in Tartu (Estonia). Mobile Tartu has taken place since 2008 and has always been an interesting place for discussions on the use of mobile data, related methods and scientific research. This special issue focuses on the use of new data in studies on mobility and transportation as well as on the methodological and operational aspects of data. Mobile Tartu 2016 was supported by the Estonian Information Technology Foundation for Education (HITSA), the Doctoral School in Economics and Innovation of University of Tartu, NECTAR Cluster 8 and COST 1305 Social Networks and Travel Behaviour.