Here’s a post about my latest significant visualisation – it’s Olympic-themed, centred around all the decathlon greats from 1984 to the present day. I’m genuinely quite happy, but not delighted, with it. Click on the image below to see the interactive version on Tableau Public. And please, if you like to explore this kind of thing, and love your decathletes, facts and figures, interact to your heart’s content!
Prior to working on this, I listened to a webinar from Andy Kirk (data visualisation expert and author, owner of which has coincided with the release of his excellent new book. I don’t need to tell you about Andy or his website, because if you’ve found my blog in the online data visualisation community, it’s an absolute certainty that you know Andy’s far more important and instructive blog already. Having also now read a significant portion of his book I can certainly recommend that too for any visualisation practitioner. Using his own visualisation project as a reference point throughout, it’s vital expert advice on all stages of preparing, formulating and executing a visualisation project. Read this post first, then go to his website and, if you’re feeling flush/generous/inspired, buy his book!
The gist of Andy’s webinar as well as his book is that there are four stages of designing a project.
          Formulating a brief
          Working with data
          Editorial thinking
          Design solution.
All these stages should be considered in order to give the best possible grounding for a successful project.
I’ve also taken a lot of stock of what Tableau’s Andy Cotgreave says, on his website. Notwithstanding the fact that the fantastic domain name makes me wonder if I should register, Andy offers great advice on iteration. His advice features the “squiggle” of data visualisation project design (below)  (which is based on Damien Newman’s Design Squiggle.)
It then features a relatively simple visualisation which had over 300 views of the data before finalising the project. The point being that the process starts with rapid exploration, and in the middle “exploring” phase (the continuing messy bit of the squiggle), we find new ideas to try, to improve on or discard as we continue the process. Only towards the end do we invest time in finessing the final product. You can read Andy’s post in full here:
So how, if at all, do these two approaches marry up – can you involve yourself in the full planning stages as suggested by Andy K, fully laying out the brief and having the solid foundations of your project ready before committing to working on it, and is this consistent with diving in to a rapid exploration “squiggle” phase of mass iteration as suggested by Andy C?
I think, to an extent, yes. The two need not be mutually exclusive, and the key difference is that in Andy C’s example (are you keeping up?) our example was a “self-commissioned” project. We have a dataset we have chosen to explore and visualise, and we are essentially our own client (although in my case, work started earlier in sourcing and shaping the dataset). However. I do believe adopting the four-stages approach of Andy K can really help. In an ideal world, the up-front preparation could really reduce the iteration at the project stage if we have a firm plan and well-formed brief. Working with the data beforehand not only ensures that we can ensure that it is clean, accurate, and in the correct format for our software package to work with, but it also allows us a sneak preview of some of the insight we might find.
So, back to my decathlon visualisation. The first questions to consider were what inspired me to create it, who were my intended audience, and what insight was I trying to show (what story was I trying to tell? – though I’m careful about using the “story word, I think this is potentially more of a “story” type visualisation than many).
I was inspired to create it because I wanted to join in with the many Olympic-themed visualisations. My favourite visualisation up to now has been my Premier League visualisation (you can find that elsewhere in my blog) which was a lot of fun and has been well-received. However the radial nature of my Premier League visualisation was not key to the insight, or the story, rather it was just an eye-catching way to display the twists and turns of a season. Described as a “polar bump chart”, I haven’t seen anyone else produce one (perhaps for good reason), so I wanted to continue along that vein. Extend this radial format (literally) to an oval shape and you have a running track, where the radial nature has more significance. Extend nine months to 32 years, substitute 38 matches for ten events, and you have the basis for a new visualisation. What I then wanted to see is how the greats of my childhood measure with the greats of recent years. My intended audience are people like me – sports fans, data fans and visualisation fans. I know that I don’t work on a worldwide newspaper or hold nearly enough fame and status to be seen by more than a handful of people (even a “viz of the day” will only take the audience from two figures into four) so the intended audience is people like me. If a hundred people like me see it and enjoy it, I’ll be happy.
We then move on to what insight do I want to display – crucially the leading message should be to see what the podium would be if all the competitors since 1984 competed simultaneously. And we do that successfully: the podium of champions features Sebrle, Eaton, Thompson. Personal delight for me that my hero from 1984 (Daley Thompson) would still be a medallist if he competed at that level today. But I found so many other stories while preparing the data and running early iterations of the visualisation. For example:
          Dmitry Karpov of Kazakhstan was the 2004 bronze medallist, but he would have been leading this fictitious all-star event well into day two. What happened to him? With no photo of him at the landing page you’d need to do a bit of exploring and moving the filter sliders to unravel that particular story.
          Similarly, Bryan Clay (2008) led after nine events but didn’t make our final podium! Was he a very poor 1500 metre runner or was his lead so strong in 2008 he could afford to relax?
          How close were all the battles in each year? In merging all together, we’ve taken that information away (visually, at least).
          Who did the best in individual events? OK you can just about make this out with judicious use of the slider, but it’s not very clear.
          Ongoing “league table” – I had a running league table available to show interim standings, which made it into so many iterations (and was thrown out of so many others). Ultimately there just wasn’t room so I went for the podium and photos instead, but I can’t help thinking there might have been a way somehow …
          What did some of the greats look like? Unless you can manoeuvre them onto the podium, you won’t see a photo. Thompson’s great 1984 rival: Jurgen Hingsen, remains consigned to the code, as I downloaded and included images for all 24 competitors, though you may never see some of them!
          Animation: I tried so hard to animate this but it just didn’t work properly. Animation meant losing all other functionality such as filtering and/or using the slider, and was just too clunky.
(the above capture from the decathlon visualisation proves that it is possible with judicious slider use to see the top three finishers in the discus, as well as to admire the photo of 1984’s Jurgen Hingsen – but you need to know where to look!)
So, I formulated a brief (display the decathlon competitions since 1984 as one simulated combined competition between medallists), worked with the data (copied, data entered and formatted in such a way that Tableau could plot all points on the curve not only for the ten events, but for points in between, so as to allow filtering at 100 points on the “track”, adopted my editorial thinking by focussing on the story, and came up with my design solution (the aforementioned unique oval racetrack bump-chart!)
My self-criticism might be me over-thinking, it might be me not making the best use of the iteration/scribble stage, or it might be that I have produced an unfinished or imperfect project. Each time I went back into the “scribble” to consider one of the points above, I either came back out with it unresolved, or made a compromise that led to a new unanswered question. I think I need to be less critical and accept there is only so much a visualisation can show. If I were to see this produced dispassionately by someone else, I like to think I would dig in, explore, find out or confirm many facts, figures and stories that interested me, and be inspired to read further (on- or offline) about the relevant competitors and events. And that’s all we can really hope for. But I think it’s best achieved by taking the time to devote to the four stages of the design project, as well as intelligent and confident use of the iteration stage.

One thought on “How should you prepare a visualisation project?

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s