Error Sources in the Analysis of Crowdsourced Spatial Tracking Data

Governments are increasingly interested in the use of crowdsourced spatial tracking data to gain information on the travel behaviour of their citizens. To improve the reliability of reporting in such mobility studies, this paper systematically analyses the propagation of errors from low level operations to high level indicators, such as the modal split and travelled distances. We find that most existing metrics in literature are insufficient to fully quantify this evolution of data quality. The propagation channels are presented schematically and a new approach to quantify the spatial data quality at the end of each processing stage is proposed. This procedure, within the context of Smart Cities, ensures that the data analytics and resulting changes in policy are sufficiently substantiated by credible and reliable information.