Week 15: Arctic Sea Ice Extent

Let me start by admitting up front that I was pretty disconnected this week. I was on holiday in Rome with my family and wanted to focus on my time there as I don’t know if I’ll ever go back. If you haven’t been, go! It’s amazing! There’s some seriously old stuff there. I’d highly recommend the Rick Steves Audio Europe app as well. It provides free audio walking tours that we do with our family. There’s a great way to see any city.

I’d like to thank Charlie Hutcheson for sitting in for me (again) for Viz Review. I really appreciate having someone to fill in so that I can make the most of the time with my family.

I was able to keep up with the posts on data.world discussion and notifications on Twitter and I was simply blown away by the sheer volume of vizzes created this week and how many of them were really high quality. Surely there were at least 150 vizzes this week (not everyone is posting on data.world for some reason). As I am going through things quickly for this review, my two lessons will be brief. I’d also like to thank Eva for her assist in coming up with the topics.

LESSON 1: WHAT MEANING DO YOUR COLOR CHOICES CONVEY?

We saw a lot of heatmaps this week, which makes sense because they show patterns very well in a small space. We also saw many different colors used in those heatmaps. Here’s an example from Arpit Arora, who has been a regular contributor lately.

The question I have is what do red and blue convey. In this example, red means heat and blue means cold. Is there anything wrong with that? No, I picked up on it straight away. However, the data set isn’t about temperature; it’s about more ice or less ice, so blue could be interpreted as water, which would indicate less ice. As proposed in Viz Review, perhaps a blue/white color palette would work better.

Using a color palette that uses common colors for water (blue) and ice (white) is an option to consider for helping the visualization fit with the theme of the topic.

LESSON 2: PROPERLY AGGREGATE THE DATA

The data provided was at the daily level. So how should you handle aggregation to higher levels like month or year? During Viz Review, Eva and Charlie reviewed this viz by Sonal Suhane:

Check out the y-axis. According to this viz, the sea ice extent is in the hundreds of million square kilometers. The largest value is the entire data set is 16.64 million square km. So how did Sonal get to the hundreds of millions? By summing the values for each day. There’s no such thing as “total” sea ice extent for a year; that’s not how it’s measured.

When looking at annual data, it would make more sense to use the median. I’ve taken Sonal’s viz and made the following changes:

Change the measure to median
Change the color palette to continuous and use the blue/white palette discussed above
Make the background grey so that the colors stand out
Make the title bigger
Change the font color to stand out well against the background

None of these changes took much time and, for me, they make the visualization more impactful.