Last week Tableau Software released their version 4.0. I love that product; in fact, I had wanted to buy it for a year before I could finally get it. I felt like a kid with a new toy. Among the various improvements in the product, this release officially makes available their great mapping function. You most probably have noticed in the last year how much using maps in data representation has become a hot topic. I guess a lot of that hype is due to all those mashups using Google Maps.
The problem with maps is that they’re in fact overused.
Visualizing data facilitates understanding and discovery. Using the capabilities of the human eyes and brain, representing information, often numbers, in visual forms, can rapidly lead to the discovery of relationships and patterns. Bear with me hear: I am actually talking about looking at stuff. Shapes and colors.
If I use maps, I expect to see new information, and make decisions about location, places.
For example, looking at this map of emergencies, I can quickly see where it would probably be a good idea to locate emergency services:
My eyes immediately see the proximity of several events; and in this case, since I need to make a decision about where I should build a firehouse and a hospital, the representation was very insightful. This is how powerful geolocation analysis using mapping can be, and you will find hundreds of much better examples out there.
Unfortunately, most often we see maps used more cosmetically, looking good rather than being really informative. A few days ago, I read a post on iPerceptions’s blog that discussed the geolocation of people doing searches about web analytics keyphrases. Here’s an example of the result map (no legend with the state names):
OK, more on the left and the right sides. I know enough about geography to call them West and East Coasts (well, this is the commonly used expressions anyway). I happen to know quite well the broad US geography, so I can recognize many of the states on this map. But there are many I can’t identify. Wouldn’t just using a plain bar chart with the name of the states be more informative?
Here’s another example of visits to my site.
OK, several among you recognized Germany. But why would I even consider this view to know which cities I get traffic from? My knowledge of German geography is very limited. Here, mapping adds nothing; doesn’t tell me what I want to know. Fortunately, Google Analytics gives us just below the table with what we need:
Frankly, I could do without the maps in GA very easily; they’re just there because they are well designed, good looking, and increase the positive perception of the application. A lot of people like them. Who wouldn’t? But as a mean of visualizing data with the goal of discovery, they’re useless.
So, be very careful next time you want to use a map to look at data. Is it really informative? Is it just gimmicks? Are you really trying to make a decision on location?
Maps are great, and their use in data analysis will have a profound impact in many fields. They’ll be more and more all over the place.
We only need to be careful not getting lost.