What telemetry data tells us
By: Sofia Heisler
Last week Buzzfeed News and the Los Angeles Times featured gorgeous street-level visualizations of how COVID-19 is affecting traffic patterns in major cities around the world. The visualizations were generated from Mapbox Traffic data.
For this post, we dug further into our telemetry data to show how much and where movement and local travel patterns have changed around the globe during the COVID-19 pandemic.
Mapbox telemetry data comprises 16 billion anonymous location points a day which we collect and aggregate for the purposes of improving the map, observing real-time traffic, and predicting traffic based on historical observations. Because we only collect telemetry from moving devices, not stationary ones, telemetry data also allows us to observe large-scale changes in movement.
Movement decreases around the globe
Below is a look at where telemetry volumes decreased throughout the world between the week of January 13th and the week of March 16th. (Only locations where we had enough data to get a reliable estimate of significant change are included.)
Trends by country and city
Here’s what that looks like for select countries, week by week, since the week of January 13th (the last week before coronavirus-related measures restricting movement were first implemented in China).
The telemetry volume in most of Europe, as well as the United States, has remained largely constant until the last several weeks, when drastic quarantine measures were implemented in many European countries and a number of major U.S. cities. We see the most drastic reduction in movement in Italy and Spain, with nearly 80% or more of traffic gone from the roads compared to the week of January 13. The US and Great Britain both show smaller decreases beginning the following week.
Here is how that looks in Europe as a time-lapse. The drastic drop the third week of March follows Italy declaring nation-wide quarantine.
Overall, the most significant changes in traffic volumes can be seen in densely populated areas. Rome leads the pack with a whopping 87% decrease in traffic, with Madrid following close behind.
Travel times in US cities:
Using historic traffic data and patterns, our traffic data is able to generate typical traffic. Here’s how travel times last week compared to historic travel times in a few US cities.
Each chart below shows the time it takes to travel a particular route from start to finish, traveling from noon on Tuesday to noon on Thursday. The pink line shows travel times calculated from “typical” traffic — i.e., how long it would normally take to travel the route on a regular day. The blue line shows travel times based on actual speeds observed on the roads on those days. Due to the decrease in the number of cars on the road, we’ve been seeing vastly improved travel times. For example, rush-hour slowdowns in Seattle have virtually disappeared, while crossing from Williamsburg to Central Park in the afternoon this week is likely to be about 25% (or about 10 minutes) faster than on a normal day.
We encourage everyone to do their part over the next few weeks to reduce the volume of our traffic data — aka, please stay home!
Live traffic data is included in directions and ETAs from Mapbox Navigation SDKs and APIs. If you’re interested in accessing global traffic data for custom routing applications or public health study, reach out to our team.
Sofia Heisler - Data Scientist - Mapbox | LinkedIn
Where and when local travel decreased from COVID-19 around the world was originally published in Points of interest on Medium, where people are continuing the conversation by highlighting and responding to this story.