Strava has released a Global Heatmap powered by Mapbox GL, plotting all locations where their users go run / go cycle / do water activities / do winter activities.
The raw input activity streams data comes from a Spark/S3/Parquet data warehouse. This data includes every one of the 3 trillion GPS points ever uploaded to Strava. Several algorithms clean up and filter this data.
The full global heatmap was built across several hundred machines in just a few hours, with a total compute cost of only a few hundred dollars.
I like the fact that it’s possible to filter on activity type, revealing some hotspots like Wielercentrum Eddy Merckx for cycling, and the adjacent Watersportbaan and Blaarmeersen for running in Gent.