An Application-Agnostic Investigation of Location-Based Services Data Quality: Using Synthetic Data and Empirical Benchmarking

Term Start:

July 1, 2025

Term End:

May 31, 2027

Budget:

$146,250

Keywords:

Data Representativeness, Disaggregate Mobility Modeling, Location-Based Services Data

Thrust Area(s):

Data Collection Mechanisms, Data Modeling and Analytic Tools

University Lead:

University of Michigan

Researcher(s):

Atiyya Shaw

Over the past decade, the widespread adoption and use of mobile services and devices have enabled the large‑scale, passive collection of human location data. The most widely available of these, termed location‑based services (LBS) data, are largely GPS enabled location “pings” collected from third party smartphone applications. These data have already proven to be useful for a wide range of mobility applications – from the joint estimation of aggregate transportation network flows (i.e., origin destination data) to examining special event and disaster evacuation movements. However, because LBS data are not representative at multiple levels (e.g., spatial/geographic, sociodemographic, behavioral), their use for individual-level (i.e., disaggregate) analyses has been limited. Nonetheless, the potential of these data for disaggregate behavior and demand modeling is a well-recognized priority area, particularly in light of growing limitations to traditionally relied upon survey data. Mobility analysis with LBS data requires preprocessing, but prior work often arbitrarily selects thresholds (for preprocessing) based on their specific end applications and unique characteristics of their vendor data. In this study, we take a methodological approach to documenting the sources of error in LBS data and developing iterative methods for evaluating preprocessing thresholds applied to this data. We propose to evaluate these decisions according to both sociodemographic and behavioral representativeness. This investigation will facilitate follow up work that entails the application of algorithms for addressing data sparsity and representativeness, and thereafter for fusing these data with travel survey data. This body of work is intended to complement existing efforts by a limited set of researchers working to advance the use of LBS data for disaggregate behavioral analysis.

Project scope

Scroll to Top