Small datasets in #MakeoverMonday tend to result in a spate of simple submissions, and that’s a good thing because I’ll take a bar chart over a maths lesson every day of the week. My submission this week was a mish-mash of indecision, but also a great opportunity to try sheet-swapping for a first time. What other submissions caught my eye this week? (at the time of writing, before Monday is even over!)

For me, the stand-out in visual terms was from Pablo Gomez. His viz is linked here, and can be seen below:

Screen Shot 2017-01-30 at 19.28.24.png

I love the look of it. Is it easier to compare dimension members than a nice straight bar? Nope. It’s bloody cool though, but as it involves curves and I know my limits, I’m not touching this one. If you download this viz from Toan Hoang you can test the grey matter with some trigonometry.

One simple submission was from Adam Crahen, and I always enjoy reinforcing some knowledge, so here’s a super quick glance at this process. First point is the calc which shows the percentage point difference between the two Measures. It could not be simpler:

screen-shot-2017-01-30-at-19-44-17

If you stick Country on Rows and this calc on Columns, you get this:

Screen Shot 2017-01-30 at 19.45.45.png

Three things to do now:

  1. Switch chart to Gantt
  2. Sort Country in descending order by [Difference]
  3. Boost the width of the Gantt line by adding a Size calc to the Size card:

screen-shot-2017-01-30-at-19-48-58

Plotting 0.15% on Size seemed about right to me, but you can do whatever feels best. Note that I also maxed the Size of Size itself:

Screen Shot 2017-01-30 at 19.50.26.png

The remaining effort is devoted to the labels. To get the + and – prefixes, the format is just:

Screen Shot 2017-01-30 at 19.52.54.png

The final notable addition in that of ASCII characters to double-encode the direction of [Difference], and it leverages a technique I saw last week from Gwilym Lookwood of the Data School. I couldn’t find nice solid arrows like Adam or Gwilym, but you get the idea:

screen-shot-2017-01-30-at-20-02-45

One oddity was that I needed to add that formatting to the Pane format to get it to stick, in spite of having set it at the level of the calculation itself. Tableau does weird stuff sometimes:

Screen Shot 2017-01-30 at 20.10.15.png

Apologies if this is a bit paint-by-numbers for many, but I still consider myself a novice and repeating exercises like above with the ASCII arrows and Size tip are still quite useful for me.

The second viz I liked was from Steve Wood. It’s another simple one, but at the time of writing it was uniquely structured in terms of week five submissions. As with Adam’s viz, I started to build this one before downloading the original. The basic structure is as below:

screen-shot-2017-01-30-at-20-39-56

Nice and easy so far. The only minor point of note is that a Group is used to split the Rows:

Screen Shot 2017-01-30 at 20.37.15.png

To do this, I just right-clicked Country, selected Create Group, highlighted United States, clicked Group and then ticked Include ‘Other’.

Back to the chart: Country is plonked on Colour to split out the Group and assign a colour to each component country in the “Other” bar. That leaves one final job: stick the label on the rightmost end, ensuring that it sums up the total at a Group level.

As is my wont, I over-complicated this. I created a dual-axis of bar and Gantt, and set the colour opacity to 0% for the Gantt:

Screen Shot 2017-01-30 at 20.51.28.png

Why do all that? Because I knew I could whack a LoD calc on the Label of the Gantt marks card to get the desired result:

Screen Shot 2017-01-30 at 20.57.18.png

I thought: “This seems a bit elaborate to do such a basic thing – is this really the best way?” The answer is a resounding no! I downloaded Steve’s workbook and was initially confused:

Screen Shot 2017-01-30 at 20.59.03.png

No dual-axis? My woefully inefficient method isn’t the best? But how the hell has he got that “label” there? What’s the SUM(Zero) all about?

Well, the SUM(Zero) eludes me, no idea if that had a purpose but nothing changed when I whipped it off the detail card. But take a look at this image – closely:

screen-shot-2017-01-30-at-21-02-53

You can see tell-tale signs of “something” at the end of each bar. What are those Gantt-esque vertical lines? Reference lines at a Cell level:

screen-shot-2017-01-30-at-21-04-55

The amount of times I forget this is frankly embarrassing! There was one thing I wanted to check before attempting to permanently store this little labelling tip. In my instance, I used a Group to separate the US from “the rest”, whilst Steve used a calculation:

screen-shot-2017-01-30-at-21-07-52

Does that different approach mean that using a Group won’t work if using a Reference Line to label the Row?

screen-shot-2017-01-30-at-21-08-58

No – it’s all good. The Reference Line sums up the sum of Employment Share along the Row, so it works the same using both methods. The LoD calc I used was utterly unnecessary.

Small tips again this week, but I’m a firm believer that the accumulation of small tips in Tableau gets you into a strong position. It’s just remembering all of them that’s the tricky part.

Advertisements