A tilemap of Switzerland cantons – with some twists

Hey, it’s Lisa again – writer here at Datawrapper. Let’s look at a map today to which the reaction of my coworker Simon Jockers was: “Wow, that’s a pretty tilemap”:

Swiss data scientist Ralph Straumann created this tilemap of Swiss cantons (the Swiss word for “states”) as a side-project in 2016 and published it through his employer EBP, an engineering and consulting company. After he added the Datawrapper-compatible Natural Earth projection to the available projections, data journalist Duc-Quang Nguyen used it for the Covid-19 dashboard of 24 Heures to show the share of people who already got a COVID vaccine in each canton.

That Covid dashboard is where I found it. Yes, this article is about Ralph’s tilemap, but I simply have to point you to that dashboard. Duc-Quang and his coworkers used all kinds of Datawrapper chart types, tables, and maps to create stunning visualizations. Here’s a selection:

Now back to Ralph’s tilemap.

What’s going on with these halved squares?

There are a few interesting things to note about Ralph’s map. He explains:

The tilemap was primarily intended for political visualizations, where the type of canton is often important. The halved squares are Halbkantone or “Kantone mit geteilter Standesstimme”, essentially half-states. They exist for historical reasons and are equal to “normal” cantons except in two regards: Unlike other cantons, they have only one representative (instead of two) in our small chamber of parliament (Ständerat). And their cantonal vote (Standesstimme) only counts half in popular votes (referenda, etc.) that require a majority of cantons/states (Ständemehr).

And what’s with these smaller rectangles in the bottom-left and top-right of the map?

They visualize the Lac Léman and the Bodensee, respectively, Switzerlands largest bodies of water. These rectangles are supposed to help with recognition. This idea was heavily influenced by the London tilemap by agency After The Flood.

How to place the tiles?

When I asked Ralph what he kept in mind when creating that tilemap, he pointed me to an article by the AirBnB data vis engineer Krist Wongsuphasawat. Using six US tilemaps for a case study, Krist defined several quality metrics for tilemaps:

  • If the overall shape looks similar
  • If regions that are neighbors in reality appear as neighbors
  • If “the relative positions between neighbor regions [are] close to reality”

Ralph, too, “tried to preserve relative locations of cantons and make the overall shape of Switzerland recognizable.”

See for yourself how he made that happen – up there are both his tilemap and the same vaccination data in the “normal” map that Datawrapper offers of Switzerland cantons.

How to create a tilemap?

When it comes to the actual software used, Ralph explains:

I created the tilemap manually in a GIS (ArcGIS, in my case, but QGIS would also work), using a background map. Pretty much similar to this figure:

I have yet to see an automatic approach for contiguous cartograms that I find convincing.

If you’re thinking about creating tilemaps and you’ve played around with R before, check out the R package called {geofacet}. Ralph has used it for other tilemaps he created: “That package comes with its own browser-based tilemap editor, which is quite helpful for tilemaps with not too many entities.”

To use Ralph’s tilemap in Datawrapper, create a new map and choose “Choropleth Map”, then look for “Switzerland > Cantons (square). To learn more about the tilemap itself, read Ralph’s explanation of this tilemap in German on ebp.ch. He also wrote about other approaches for cartogram creation on his blog. We’ll see you next week!