Last month, over 200 million viewers watched the 2016 Eurovision Song Contest. The annual musical extravaganza features a singing competition, in which each European country sends a homegrown singer to the live performance. Meanwhile, viewers vote on the year’s most popular performance and tune.
Throughout the telecast, viewers also turned to Shazam to help identify the song they’re hearing. Hear a catchy song and wondering who sings it? Use the app to record a snippet and identify it. Shazam received some 262,000 requests over the course of the broadcast, and came to us to see what we can glean from the big dataset.
Technical considerations
Datasets of this size are often tricky to visualize, but vector tiles rendered with Mapbox GL provide two distinct advantages:
Using a tool like Tippecanoe can simplify geometry at lower zoom levels, reducing the rendering load to a manageable size without changing the overall distribution of points.
The data that is loaded on the map is held in memory, enabling us to query them instantaneously in the browser without a separate request to the server.
Surprise findings
While some UK citizens may have voted to leave the European Union, Shazam’s data shows that Brits want to continue to be part of Eurovision. With more than 5,300 requests to Shazam, Britain was as engaged as Austria and Italy, but not as Euro (and music) excited as Sweden with more than 30,000 requests.
In the map below, we’ve filtered the dataset by geography: hover over each country, to see where viewers wanted to identify contestants’ tunes.
Visualizing change over time
Mapbox GL can filter on any attribute, and we can make a time series map by segmenting the queries by their timestamps. Move the slider to explore clusters around major cities, and see engagement spike across nations over the course of the telecast.
Visualizations like these are just some of what you can build with Mapbox GL. We can’t wait to see what else you come up with!