A Statistical Approach to Account for Missing Data in GPS Trajectories

• Activity Number: 177 - Observations Through a Foggy Lens: Modeling Complex Measurement Error and Non-Random Missingness in Ecological and Environmental Health Data
• Type: Invited
• Date/Time: Monday, August 8, 2022 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Consulting
• Author(s): Marcin Jurek* and Kate Calder and Corwin Zigler
• Companies: University of Texas at Austin and University of Texas at Austin and University of Texas at Austin
• Keywords: spatio-temporal statistics; human mobility; trajectory imputation; missing data; GPS data; digital phenotyping

Abstract

The advent of smartphones opened new horizons for statisticians by producing new types of data and in unprecedented amounts. In particular, it allowed recording the user's geographical location using GPS which enabled investigation into human mobility patterns with more accuracy than ever before. Unfortunately, in practice, individual GPS tracks often have large gaps which present a challenge for researchers trying to use this type of data. While some methods have been developed to address this problem, they are often very simplistic or ad hoc. We propose a statistical model to represent human mobility at an individual level, we use it to impute missing parts of the trajectories and we demonstrate how our model can be used to quantify uncertainty.