Context
Recently, I have noticed a substantial increase in the use of tile maps to represent geographical regions, particularly ones featuring the States of America. Over and above looking quite stylish and futuristic, the hexagons can sometimes (not always) improve the story that a dataset has to offer. This was particularly true of a Makeover Monday a few weeks ago, focusing on US Household income, where tile maps seemed to be the most popular method for displaying the data.
Having seen so many US Tile maps, I thought it was fine time to try and reproduce one for the UK. Using counties instead of states, I set about drawing up a template for our green and pleasant land. The following methodology has been designed to be replicable for any country or group of countries (it may be interesting seeing a continent in this fashion rather than just a country).
Methodology
I had considerable difficulty trying to position counties in rows and columns that were geographically similar to their actual location, while trying to maintain the general outline of the United Kingdom that we are familiar with! Take Edinburgh & Glasgow, for example; in reality they occupy a very similar latitude. However, due to a congestion of small counties in this part of Scotland, I had to bump Edinburgh up a row to save Scotland from looking too wide. After much trial and error I stopped playing and settled on a final hexagonised UK, although I’m still not overjoyed with my Cornwall or North Wales. The screenshot below shows my final mock up of the UK in Powerpoint – by tracing hexagons on top of a map of the counties, I was able to create a grid of rows and columns which laid the basis to the Tableau version of my map.
Starting in the bottom left hand corner, (coordinates 1,1), I then gave each county it’s own row and column coordinates. (Note that values for columns in ODD rows have been pushed across by +0.5 to create the interlocking effect).
As in the US tile maps, I also created an abbreviation to sit on top of each hexagon, though I realised that these are only partly useful – they still require a reasonable knowledge of the geography of the UK to be interpreted, unlike in the US where two letter codes are used far more often to represent states. CA continues to be synonymous with California, not Cambridgeshire or Carmarthenshire!
It was then time to draw my map up in Tableau and apply some data to bring it to life. I have included several tutorials that I found useful in an appendix, but in essence it is simply a case of adding your rows and columns variable to the appropriate shelf and then playing around with shape & size to get to your map. I chose two very easily accessible and understandable measures (population & area) to illustrate the maps and combined them to make a third (population density). I have included a section for further potential analysis in part 2.
Results
The image below shows the results of my first two hex maps. Immediately, I noticed a key issue with my maps: the data. Namely, the Highlands in the North and London in the South. These anomalies skewed the colour scheme so greatly that smaller counties became lost when making comparisons.
In order to rectify this, I had to manually set a maximum for the upper bound of the colour scheme. I did these for each of my three measures and it brought the story into clearer focus. Though this particular dataset does not offer any particularly new insights, it becomes obvious through the maps which counties have the highest populations: the South-East with London and the its Commuter counties and the North West of England with its large cities of Manchester, Sheffield, Liverpool & Leeds etc. The Area map illustrates the vast spaces in the Highlands of Scotland, Yorkshire, Lincolnshire & Devon. Population Density, meanwhile, highlights those counties that are effectively cities that have grown to absorb surrounding villages and towns (Cities of Glasgow & Edinburgh, Greater Manchester, Bristol & Greater London).
Conclusions
I set out to build a UK Counties Hex Tile map and to that extent have been successful. My next question surrounds whether they are an effective way of showing data. I would argue yes to this, although it depends on the data the user is trying to show. In the above examples the maps were initially limited by the large values found in London & the Highlands of Scotland, though with a manual fix I was able to draw out more information on this occasion, the trade off being that it becomes less obvious just how much bigger London & the Highlands are in their respective measures. This led me to think that the Hex map would be particularly useful for showing relative measures rather than absolute measures – this is something I’ll look at in part 2 of this piece.
Resources
Here is a link to the Tableau Viz on Tableau Public:
Below is a list of blogs that I found useful as a starting point for building my own hex map.
http://www.sirvizalot.com/2015/11/hex-tile-maps-in-tableau.html
https://www.tableau.com/about/blog/2017/1/viz-whiz-hex-tile-maps-64713
https://www.tableau.com/about/blog/2016/7/how-create-density-maps-using-hexbins-tableau-56511
Comments