Human mobility research / Research software
NOMAD
Contributions to an NSF-funded platform for processing large-scale GPS mobility data, validating stop-detection methods, and studying co-location.
Dashboard preview
Stop-detection algorithms in motion
A recorded run of the public dashboard, which visualizes synthetic mobility trajectories and compares stop-detection output.
Problem
Human mobility studies often depend on large, sparse GPS datasets and inconsistent preprocessing, which makes methods difficult to compare, scale, and reproduce.
Approach
At Penn CSS, I contributed to NOMAD's open-source Python/PySpark library and stop-detection dashboard: contact estimation and social interaction potential, temporal-blocking performance work, edge-case tests and notebooks, synthetic-oracle validation, and dashboard interactions and static deployment.
Result
The work supports reusable co-location analysis, faster radius-based contact queries, more robust validation workflows, and a public browser-based demonstration of stop-detection algorithms.