In the last of this series, I looked at a nice viz from Lindsey Poulter. In that post, I cited David Krupp as another inspiration of mine, and it seems fitting therefore to deconstruct one of his recently shared visualisations.
I’m like a stuck record, but this is yet another example of the benefits of the #MakeoverMonday initiative. Even if you don’t learn directly by participating, you can still learn indirectly by observing what other people in the community are doing. This second strand is vital for me, as it helps to shoehorn a bit of “design” into my head, and it’s important as it is not a simple thing to just pick up and run with.
A lot of Tableau training inevitably focuses on technical aspects of the tool, but it’s arguably the design element that most effectively transmits the message. It has been interesting to see prominent Tableau social presences either always champion the KISS message, or to revert to that position, such as Rody Zakovich did, supported by this excellent post.
Even more recent than Rody’s post is the video below, which has been uploaded following the Tapestry Conference. Cole Nussbaumer Knaflic doesn’t need introduction, and it’s the combination of the KISS message and the art of storytelling with data which I am trying to take on board with my own learnings.
But that’s one Hell of a digression. Let’s look at David’s latest viz. As with Lindsey’s, I’m just downloading the workbook and checking out each individual worksheet. At first glance, I’m assuming that we have four worksheets for the four charts, and then a couple of text boxes in the body of the viz, with a further text box acting as a title at the top. I also assumed it to be a tiled rather than floated. I was perhaps half right.
And we’re off! A straightforward opening, with the exception of the fact that David has elected to aggregate the ratings:
My immediate thought was: “Why bother to create a calculation, couldn’t you just create a Group?”:
This Group already existed in the workbook, so David may have chosen to switch approach part way through the viz. At this stage, it looks like both yield the same end result. Let’s take a glance at the logic behind the colour calculation:
As above, is there an argument that you could just use a Group in a calc to get the same result? You can’t:
Sticking that on Colour is a little underwhelming:
It’s a whiteout! You can’t use Groups in calculated fields. Not sure why. It would seem a sensible thing to be able to do, but you instead need to create groups via calculations as David did, or you could create a Set and then reference that in a calculation. One for the wishlist.
Sheet two looks largely the same. There’s a different metric, but the same structure and the only subtle variation is that this sheet has shading applied to differentiate it from the prior worksheet, which is plotted on the same horizontal plane in the end dashboard:
Two down, two to go. Time to look at the third sheet, which represents a departure from the bars:
OK. So we’re plotting the highest ranked Social Blade Score in the middle of a circle. Why is it dual-axis? Because David wanted to label the Rating Group under the circles, and the ranking inside the circle. You can’t do that if both are lobbed onto one label, so you do one on one axis, and the other on another. Curiously, I noticed that David had put a copy of his Rating Groups calc onto the Label shelf:
Look what happens if it is the original Ratings Group calc:
Random space. The calcs are exactly the same. I have no idea why this has happened and I’m not going to dwell here either! One minor point to note is that the axis is reversed, to ensure that the lower rank is rightly represented as the “better” end of the scale:
The final chart is back in bar territory, this time showing the number of channels within each Rating Group. It’s essentially using the same technical principles as sheets one and two, so I’ll move onto the construction of the dashboard itself.
It is not tiled!
The first thing that caught my eye was the title, and I hadn’t really considered how it would be built in Tableau. It transpires that it wasn’t: both the main title and the tagline are Floated images. I guess they were knocked up in some other software package:
Both of the Objects prefixed with “SocialAnalytics” are “title” elements. The black horizontal and vertical bars throughout the dashboard are simply shaded and sized text boxes. Larger, shaded text boxes act as backgrounds for the six Objects in the body of the dashboard:
Where charts are floated on top of these backgrounds, you can see the reasoning behind shading the backgrounds in some of the worksheets:
Some of the floated Objects are simple text boxes, which were needed to sit over the standard axis labels as in the above image, to make the “A+ or A” red:
Being able to colour Discrete axes based on calculations would be pretty cool, wouldn’t it? Another one for that wishlist.
And these things add up. There are thirty separate Objects floated with precision on the custom sized dashboard. Each of the titles and annotations in each “pane” is actually an image:
Even the “text panes” are images. David must really like that font! What do I ultimately take away from this exercise? Things are not always as simple as they look! None of the “technical” aspects of this visualisation in a Tableau sense are beyond the capabilities of a novice user. However, the simplicity of the aesthete of the end product belies the amount of work involved in creating such a clean-looking dashboard.
When I look at anything that David, Lindsey or Chantilly Jaggernauth create, there is often this impressive amount of “unseen work” that goes on when creating such crisp visualisations. It’s well worth downloading and exploring any of their vizzes, as it’s an inspiring and eye-opening experience.