Quantcast
Channel: maps for developers - Medium
Viewing all articles
Browse latest Browse all 2230

Shopping and Sunbathing: How America Responds to Quarantine

$
0
0

By: Scott Farley

Chances are, you’ve probably waited in a longer-than-normal line at grocery stores recently. Perhaps you’ve jogged past a few more people in your neighborhood park lately. And, as only essential businesses remain open in most parts of the world, you probably haven’t spent a lot of time in your favorite coffee shop, dined in your favorite restaurant, or visited your favorite bookstore in the last few weeks. All over the world, people are interacting differently with the businesses and public facilities in their communities.

We used Mapbox Movement data to look at where people are spending their time, and where they’re not, comparing distributions of activity in April to those in January. This rich dataset, composed of anonymized and aggregated location updates from mobile phones, reveals representative movement patterns of millions of users worldwide.

We looked at a number of commercial venue types (grocery stores, travel agencies, restaurants, bars, etc..) and a few public facilities (parks, beaches, swimming pools). We then looked at the relative change in visits to these locations in several cities across the country. We found that, while there have been large declines in activity across the board, there are differences in both categories and geographies.

Percent change of visits to various categories between January and April, 2020 on the horizontal axis. The total number of times a place of that category was visited is shown on the vertical axis. Circles are sized by the total number of venues in that category. Data is from activity patterns in the Los Angeles, San Francisco, Seattle, New York, Atlanta, Orlando, Chicago, and New Orleans metropolitan areas.

Perhaps not surprisingly, theme parks, schools and gyms experienced the largest decline, along with bars and movie theaters. But there are other interesting patterns presented as well, such as:

  • Beaches experienced a small positive increase in traffic as the weather warmed. Beaches have also been at the forefront of policy debates around the closures of public spaces.
  • Activity on trails and golf courses declined only slightly — these areas have largely remained open for general use and solitary recreation.
  • Grocery stores across all geographies saw approximately an 30% decline in visits between January and April. Along with pharmacies, many grocery stores saw small peaks just before stay at home orders were enforced.
  • While restaurants continued to have a significant number of visitors (for example, by those ordering take-out or curbside pickup), traffic to these venues declined by almost half.

While these patterns are interesting in aggregate, they’re more informative when further broken out by geography. For example, consider food service venues (in this case grocery stores and restaurants) in 15 populous counties. Without exception, visits to restaurants in these counties declined more than those to grocery stores — which makes sense, as many restaurants have been closed under various shelter-in-place orders. But the difference in decline between the two categories is not homogenous. For example, visits to restaurants in Fulton County, GA, declined almost 75% between January and April, but visits to grocery stores were down just half of that. Similar patterns are visible in other counties like Alameda and Santa Cruz, in California. New York County, NY, on the other hand, saw major and nearly equal declines in grocery store and restaurant visits, perhaps indicating that residents turned to alternative means of obtaining food, such as online grocery delivery.

Percent change in grocery store and restaurant activity between January and April

In all counties, food-related activities like grocery shopping and restaurant visits are down significantly. In some counties, however, the rate of recreational activities like hiking and going to the beach increased substantially between January and April. For example, beaches in King County, WA (Seattle area) experienced an 11% increase in visits during this period. Middlesex, MA (Boston area), Suffolk, NY, and Bergen, NJ all had significant increases in beach visits as well.

Change in beach visits, January to April

Per-county aggregations also expose temporal patterns in visits to essential and soon-to-be essential businesses. The Bay Area shelter in place order was enforced starting on Monday, March 16th. In the several days leading up to the closure, there was a large spike in activity at pharmacies and other essential businesses as people waited in line to buy food, medicine, and other household items. Alameda, Santa Clara, and San Francisco counties all saw higher than normal traffic to pharmacies during the weekend leading up the March 16th closure, after which traffic to these venues declined substantially.

Visits to pharmacies in Bay Area counties, compared to the January baseline

These aggregate differences in activities are more interesting still when they reveal patterns about the communities they represent.

For example, in New York, the distribution of grocery store visits is heterogeneous in time and space. On a single day, Saturday, April 5th, most neighborhoods in the New York metro area had a relatively low number of grocery store visits compared to the January baseline, aligning with the pattern of the city overall. Some areas, though, like Park Slope, in Brooklyn, had an increase in grocery shopping activity on this day.

Change in grocery store visits, January to April, aggregated by zip code for Saturday April 5th, 2020. Darker brown indicates significant negative change (fewer visits). Teal indicates positive change (more visits).

Similarly, in the Los Angeles area, many neighborhoods experienced small to moderate increases in grocery store activity compared to the baseline, particularly on the weekends.

Change in grocery store visits, January to April, aggregated by zip code for Saturday April 5th, 2020. Darker brown indicates significant negative change (fewer visits). Teal indicates positive change (more visits).

Visits to parks in Los Angeles further emphasize the spatial heterogeneity of the city. Neighborhoods like Beverly Hills, Santa Monica, and Burbank all had over 50% fewer visits to parks in April than they did in January. Other parts of the city, however, experienced increases in visits to parks during the same period of time.

Visits to Parks in Los Angeles County during the month of April. Again brown indicates significant negative change (fewer than normal visits). Blues indicates neutral or positive change compared to the January baseline.

This is only the tip of the iceberg. There are many more interesting spatiotemporal patterns we can dig into, using activity data as a proxy for behavioral patterns more generally. The Mapbox Movement dataset is continually updated, allowing us to see how patterns are changing through space in near real-time.

To learn more about Mapbox’s Movement data product, take a look at our Data Products Page or check out other COVID-19 related mapping projects using Mapbox.

Scott Farley - Senior Data Scientist - Mapbox | LinkedIn

Maps feature data from Mapbox and OpenStreetMap and their data partners.


Shopping and Sunbathing: How America Responds to Quarantine was originally published in maps for developers on Medium, where people are continuing the conversation by highlighting and responding to this story.


Viewing all articles
Browse latest Browse all 2230

Trending Articles