As some of you may know, I’ve recently started teaching data visualization at the University of Utah. I’ve only been at it a couple of weeks, but it has been a great learning experience for me (and hopefully my students). Now that I am teaching the subject, it seemed like a decent time to do some self reflection to see what I might do differently in some of my past Tableau Public projects.
Much (exactly) like Makeover Monday, this series will take a look at a visualization (of my own) that is in need of improvement, with the end result being a new improved visualization that clarifies the original message. Customary to Andy’s original format, I’ll also be describing things that work well/didn’t work well with the original visualization.
For the second entry to this series, I’ll be looking at my most popular Tableau Public viz: #IronViz Wikibirths. This viz was created in March 2015, and was my first venture into the Iron Viz ring. This viz was a fun project for me, as it also started my interest in scraping web data with Python. The dataset used for my entry was harvested across 366 Wikipedia pages (like this one) in only minutes.. it was then that I started to understand the power of having additional dev skills outside of Tableau.
My viz didn’t win a trip to Vegas… Shine showcased his amazing storytelling skills and went on to win the main competition later that year. Somewhere along the way, this crazy orange viz of mine went viral, and somehow has over 200,000 views to date! After taking a careful look back, I don’t think all those views are really warranted, as the viz itself is pretty confusing. Join me, and guest bloggers Pooja Gandhi and Adam Crahen of the Data Duo, as we critique and makeover this orange mess.
I will never not be amazed at the amount of views this viz has, so I must have been doing something right, but let’s turn the time over to the Data Duo to see what actually works and doesn’t work. I’m guessing there will be more that doesn’t work…
We chat with Curtis all the time and have become great friends over the last year and a half. We often share work amongst each other and ask for feedback. One day, during a recent conversation, he casually slips in this request…
We both eagerly agreed. Only realizing minutes later that he just asked us to critique a single viz that has more views than probably all our 200+ public visualizations put together. No pressure at all so here goes.
What works well?
What doesn’t work well?
Overall, this is a great viz and is worthy of the views it received. However, we know Curtis has improved over time and we cannot wait to see his makeover. It will probably get another 200 thousand more views.
With Adam and Pooja’s thoughts in mind, let’s break down how we can
trash improve on the original. One key concept I’ve tried to latch onto is the thought that if you visualization still tells the same story in grayscale, then your color and design choices are probably fine.
Let’s take a look at what happens if we push all the colors to the background…
Pushing to gray really highlights one of the Duo’s comments: Five charts encode the same color differently. You could imply that a darker shade of orange/gray represents more significant births, but the coloring is still incorrect… one dark orange does not mean the same thing as another dark orange, and that fact isn’t readily apparent to the audience. The area chart is also in shades of orange, but has nothing to do with volume of births, which simply adds to the confusion.
After pushing the viz to grayscale, I went through a few of the different elements after finally landing on a conclusion.
I’ll be honest, throwing this area chart in the trash was a tough decision as I think a lot of people immediately recognize that as my work. I went back and forth on a decision to remove or improve..
I thought I had a decent solution to the coloring problem (below), but the stacked area chart was still just for show, and didn’t tie back to the original question in the title.
You’re still here? This got a bit wordy, so I appreciate you sticking around until the end. Here is the new visualization, and some thoughts as to why it is a major improvement to the original.
Overall, I chose this viz as an exercise in user interaction and color. I was going for high impact in 2015, and consistently trying to do things in Tableau that were out of the common realm… looking back I’m seeing some of the trash that mindset let me to produce. I’m hopeful that the new visualization can hold its own against its orange brother… I’m afraid blue is going to be looking up at those view counts for a long time to come!