Scaling Behavior of Human Mobility Distributions

What:
Presentation
When:
20 minutes
Tags:
spatial datamobilityGPSanalytics
Discussion:
0
Recent technical advances have made high-fidelity tracking of populations possible. However, these datasets, such as GPS traces, can be comprised of millions of records, well beyond what even a skilled analyst can digest. To facilitate human analysis, these records are often expressed as aggregate distributions capturing behaviors of interest. While these aggregate distributions can provide substantial insight, the spatio-temporal resolution at which they are captured can impact the shape of the resulting distribution. We present an analysis of five spatial datasets, and codify the impact of rebinning the data at different spatio-temporal resolutions. We find that all aggregate metrics considered are affected by rebinning, but that some distributions do so regularly and predictably, while others do not. This work provides important insight into which metrics can be used to compare human behavior across datasets and the kinds of relationships between that can be expected.
Participant
University of Saskatchewan
Participant
University of Saskatchewan
Participant
University of Saskatchewan

My Schedule

Add to Your Schedule