|
Abstract:
|
Large-scale social distancing measures have been a critical part of the global response to COVID-19. In order for public health experts to understand the effectiveness of these measures over time, many have turned to aggregated mobility datasets collected from cell-phone and app users. Facebook’s Movement Range Maps are one such dataset. This dataset, based on data from Facebook users that have opted-in to sharing their location data, includes two distinct metrics that capture how population-level mobility has changed throughout the pandemic; each metric aggregated to geographic regions comparable to U.S. counties. One metric, called Travel Range, is a measure of how much people are moving around compared to a pre-pandemic baseline. Another metric, called Stay Put, shows what fraction of the population is restricting travel to a small area around their likely home. Since location data is sensitive, it is imperative that these datasets employ robust measures to protect privacy. In this talk, we’ll describe how these datasets are constructed and how we employ differential privacy to protect the privacy of Facebook users.
|
