Duplication in Design

At the end of November 2018 I created a visualisation detailing all 140 space shuttle flights, from the first non-captive flight of Enterprise in 1977, to the final flight of Atlantis in 2011. Anything space related has always fascinated me and I’ve never forgotten watching a space shuttle launch in 2002 (I just so happened to be in Florida on holiday with my family at just the right time!)

I spent a good bit of time of this viz and was really pleased with how it turned out. It got a nice reception on Twitter, and it even got picked as visualisation of the day on Tableau Public!

 

Click the image for the interactive version

 

I’d decided on the basic layout fairly quickly along with the charts themselves. The line charts and unit charts seemed to fit the data quite well, and they aren’t overly complicated to put together: a running total, the index function, and bins (if you’ve got any questions on how to build the charts find me on twitter). However the majority of my time was spent on the design and fine tuning the layout.

 

Duplication is Good

Below is a simplified diagram of the viz showing how each of the six shuttles has its own section on the dashboard. Mocking something like this up in Power Point or on a bit of paper is a great way to get the design process going quickly. Almost all my visualisations start off with something like this.

Template 01.png
 

Each of the six shuttle sections then has its own set of three charts

 
123.png

  1. Running total of the number of launches (Chart 1_)

  2. The individual flights in the order they launched (Chart 2_)

  3. Duration of each flight, ordered from shortest to longest (Chart 3_)

 

Adding those three different chart types to the diagram, you can straight away see a significant amount of duplication between the six sections.

Template 02.png

With six versions of each chart, each one showing specific information for that shuttle, this makes for a total of 18 charts in a relatively confined space.

In theory I could have designed the visualisation in a far more compact way, just using the three chart types and provided a filter to select the space shuttle you wanted to view (see below). This makes a nice and succinct visualisation that gives a good summary of the selected shuttle. But that’s where this sort of design can become limiting, and comparisons are more difficult as you can’t see all of the data.

Duplication SSO-Main.png

So repeating the charts for each shuttle, and being able to see every single flight at the same time, really helps with comparisons between them. You can see which Shuttle had the most flights (Discovery), how frequent the flights were (steeper line = more regular flights), and the number of flights by duration (8 of Columbia’s flights were 15 days or more).

 
 


It’s very easy to argue that the line charts are potentially redundant. Instead of showing a single chart with 6 lines (with one for each Shuttle), or even 6 charts with a single line each, I’ve shown 36 lines (six for each shuttle). Writing it like that sounds pretty bad and fairly pointless. But highlighting one shuttle in each of the line charts really makes the data catch your eye. I feel that the repetition, in conjunction with the highlighting in white, helps reveal patterns and connections in the data that might not be obvious at first glance on a single chart.

 

Duplication is Bad

So I’ve talked about why duplication can be good, it can also be a huge waste of space.

Don’t duplicate an axis where you don’t need to. As the line charts are positioned in two horizontal groups, I only placed a y-axis on one side of each group of three charts. If they’d been grouped vertically, I’d have done the same for the x-axis.

Trust your audience and don’t label everything. The unit charts on this viz all have the same layout and show the same aspect of the data for each Shuttle. So I only put titles and axis on the first of the six charts, as my audience can refer back the first chart if they need to.

And talking about chart axes, if you see one with labels like 1980, 1990, 2000, 2010, you don’t need a label saying years, it should be obvious. And if you’ve got a chart title that states what the chart is showing, that can stop the need for axis labels as well, as you’d potentially be duplicating information for little gain.

 

Use Cases

There’s nothing to stop this sort of design being used in a business setting. In the case of this visualisation you could almost do a direct swap:

Shuttles = Departments/Categories/Locations

Launches = Sales/Customers/Products

I’ve lost count of the number of dashboards I’ve designed that end up being printed out or just used as a static image within a Power Point presentation. And this isn’t necessarily a bad thing, provided someone can still get the insights they’re after. So a design like this can help an end user get more insight when a dashboard isn’t being used interactively.

 
 

I’m doing loads of baking at the moment, here’s one of my more recent attempts.

Use this recipe, it’s amazing

IMG_7699.JPG