Today, Andy Kriebel tweeted a viz that I liked the look of, so inevitably I downloaded it to see how he achieved the things he did, and I picked up a few tricks on the way. The viz in question is linked here, and looks like this:

screen-shot-2016-12-03-at-20-27-46

What do I like? It’s clean and ticks a number of design boxes:

  • Legend up at the top, so you read and understand that before even reaching the charts
  • Strip plot at the bottom confirms Year, and is aligned to the bars above – great use of space to have just one Time axis covering two charts
  • I really like the Summary on the right. It’s simple and gives a punchy overview of what the 114 years of the Bermuda Cup boils down to
  • The typeface caught my eye – I’d used American Typewriter a week or two back for #MakeoverMonday, so it was nice to see it again

I’m not even going to recreate this viz, it was just a case of downloading the workbook to examine each of the worksheets to understand how it all hangs together. First up was the top chart:

screen-shot-2016-12-03-at-20-34-05

Structurally, there are no real surprises. The x-axis was obviously a time series, and I guess the only debatable decision was that of selecting a Continuous rather than Discrete time series. What is the impact of switching to Discrete?

screen-shot-2016-12-03-at-20-36-18

A straight switch to a Discrete does the above, but it can be remedied by tweaking the Size of the bars and circles. It’s no big deal, and things seem right off the bat with Continuous, so that seems to be the deciding factor.

Let’s look at the calculations.

screen-shot-2016-12-03-at-20-38-49

Nice and easy. If it’s one team, pull through the Margin of Victory Measure, otherwise invert that same Measure. Hang on! Surely the y-axis should reflect that? How has Andy got positive values showing for each team? Check out the default number format of the calculation:

screen-shot-2016-12-03-at-20-40-35

Sneaky and effective! Irrespective of whether the Measure is positive or negative, it will be depicted as positive. It’s one of those classic, subtle Tableau “hacks” which achieves a nice end result.

Now let’s look at the other calc:

screen-shot-2016-12-03-at-20-42-32

Again, nice and easy. The core of the viz is depicting the winning team and their Margin of Victory. Where the result is a draw, it makes complete sense to have a dual axis on which a tied result is shown in a neutral position, and as a different type of chart (Circles in this instance). OK, I understand one component of the dashboard. Time to look at the Strip Plot / Barcode.

Screen Shot 2016-12-03 at 20.45.45.png

Immediately, I don’t understand two of the pills. What is “”, and what is the 0.5 Measure all about? In the first case, I tweeted Andy himself to understand the purpose of the weird “” on Rows. I could see that if I removed it, this happened:

screen-shot-2016-12-03-at-20-47-12

See how the white space to the left of the original has been lost?

screen-shot-2016-12-03-at-20-48-14

As ever, Andy was really quick to respond to the question, and that’s just symptomatic of the kind of help I’ve been receiving from the Tableau community in the past 9 months or so. It’s a huge help, and I appreciate everyone’s time and patience when fielding all of my questions!

His answer makes sense when you consider the alignment of the charts on his final dashboard. Without the white space, the time series’ of the bar and strip plot charts would have been misaligned. By forcing the alignment, he could use the strip plot as the only labelled time axis for both charts.

Andy also provided the bonus answer of what the 0.5 was for. Look what happens if I remove it:

screen-shot-2016-12-03-at-20-51-39

What about if I tried to set the chart type to Gantt?

screen-shot-2016-12-03-at-20-52-47

OK, it makes sense now. That crazy amount of white space above and below the Gantt isn’t great, which is why using bars was needed. However, without having a value to plot, Tableau drew a line, hence the need to force in a uniform Measure to draw the Bars.

The final interesting element for me was the Summary sheet:

screen-shot-2016-12-03-at-20-54-52

For this viz, it is a critical addition as it gives an at-a-glance Summary of the history of the Bermuda Cup. The Bar and Strip Plots show which team won, and in the case of the Bar – by how much, but you can’t use those charts to determine who has won most matches overall.

In this case, I don’t really understand the point of the Column Measure. What happens if I remove it?

screen-shot-2016-12-03-at-20-57-50

Everything right aligns as it has no horizontal reference to anchor to. OK, so what if I just shove a “1” on Columns?

screen-shot-2016-12-03-at-21-02-02

The Winners get staggered, as the number of wins / draws is plotted against a variable, continuous axis. What if I change the aggregation of “1” to an average:

Screen Shot 2016-12-03 at 21.03.32.png

This time, everything is aligned against “1”, but the axis range means that it is overall aligned to the right. When you plot against the Average of zero, there simply is no axis range, so everything ends up centre aligned.

I learn a lot from downloading and breaking down workbooks, and I can’t see that I’ll ever stop doing this as I just don’t believe there is a point at which you stop learning things about Tableau – certainly not after 8/9 months! Hopefully this post reveals one or two tips to anyone reading that they weren’t 100% sure of.

Advertisements