I’m going to start with a complaint. There is an assumption that if you submit your work for Viz Review, you will iterate on it. Far too many people received feedback and didn’t iterate. In fact, most didn’t. You are taking time and feedback away from people who we are not getting to. Please be respectful of them.
This week we looked at a selection of cities across America and how long it takes people to pay off a mortgage based on the median income and median home price. It was great to see the original creators, howmuch.net, engaging with the Makeover Monday community.
As I’m on holiday, I’m going to keep this recap brief so that I can relax.
LESSON 1: TREEMAPS REPRESENT PARTS TO WHOLE
What are treemaps for? Treemaps display hierarchical data in a single view, split up into a series of rectangles. The size and position of each rectangle is determined by the proportion each dimension represents as a part of the total.
The data this week DID NOT represent every city in the United States, or even a single State. Therefore, using a treemap, or any other forms of parts-to-whole, is misrepresenting the data and a poor form of visual communication. Consider this example that Chithresh Suresh submitted for Viz Review:
Presenting the data this way only work if every single city in the United States is in the data set. On top of that, the data represented in the average number of work hours that people need to work in each city. How can you even add those up? Go with a bar chart.
LESSON 2: REPRESENTING A SUBSET OF DATA AS THE TOTAL
Consider this example submitted for Viz Review by Zunaira Rasheed:
There are two big problems here:
- By using a hex map, Zunaira is representing each State as the sum of the cities within each State. This is wrong. The data is a subset of cities and there are not cities in all States in the data set anyway, thus leading to missing States on the hex map.
- The note on the right say that it takes people in California 1,157 hours per month to pay off their mortgage. That’s 29 weeks. The last time I checked, there aren’t 29 weeks in a month. You CANNOT add the data together.
Now here are this week’s favorites.