COVID-19 Mobility Trends Part 2: Visualizing Google's Data

Two days ago, I published a new blog post where I broke down mobility data that Apple had generously made public to aid cities, researchers, and the public at large. One of the things I noted in my post was that while the granularity of Apple’s data trailed what Google had previously shared, Apple made their data easily accessible via CSV download, while Google’s data (though seemingly great) was locked up in PDF files.

Well - what a difference a few hours makes! Since my post (not saying it’s causation, but 🤷), Google has added a “Download Global CSV” option to their Community Mobility Reports site.

With the ability to denote which kinds of places activity is occurring (residential, workplace, grocery, etc.), Google’s dataset captures a bit more nuance than what Apple has shared. As such, I was eager to to spend a bit of time playing with the data! I wasn’t expecting to reach any conclusions materially different from what Apple’s information yielded, but was curious about how the increased visibility into place type might inform my understanding of the world’s “new normal”.


Timeseries of Activity by Place Type

Changes in Activity as of 4/11

Mapping Activity in Parks


  • As expected, activity tagged as “Residential” is up significantly as shelter-in-place orders and enforced work-from-home policies contribute to people holing up at home
  • Google’s data re-emphasizes one of the key findings from Apple’s mobility data, which is that transit usage has absolutely dropped off a cliff since the start of March
  • In addition to transit, Apple’s data told us that folks are driving much less… Google’s data is useful because it also tells us where they’re not driving to, with Retail/Recreation and Workplace experiencing large drops in activity
  • While Grocery/Pharmacy and Parks also fall off their February baselines in a significant way, they nevertheless see an activity decline only half as much as Retail/Recreation
  • Park Activity interestingly has the broadest distribution of activity of any place type, with most countries seeing a material decline in activity, but a not-insignificant number of outliers with more or significantly more park activity than February
    • To see exactly where, I put together a map of park activity which clearly shows that Northern Europe, with Sweden and Finland in particular, is the source of those outliers
    • Much has already been written about how these countries are choosing to control the pandemic - it will be fascinating to look back in the future and see whether or not the looser model of pandemic response these countries have chosen was the right decision

Google Data Attribution

Google LLC “Google COVID-19 Community Mobility Reports.” Accessed: 2020/04/18.