So, what were the chances of that?!

So, what were the chances of that?!

This weekend (I’m writing this on a Tuesday evening) there’s the deadline for entering the second feeder contest for Iron Viz. Iron Viz is one of the highlights of every Tableau Conference, where three winners of the feeder contests perform on stage in front of 12000 or so people to produce an amazing viz in just 20 minutes. Possibly the most ridiculous display of nerve and talent you’ll ever see in the world of Tableau, a great spectacle that leaves those watching in awe of the three competitors.

To take part on stage would be absolutely out of my league, but there are still good reasons to enter. Firstly, the task of pushing myself to produce a competition-quality visualisation that wouldn’t seem too out of place in a field of 60 or so competitors will be a great experience. And secondly, if it was possible, just possible, to win the feeder, it would be a free ticket to this year’s Tableau Conference in Vegas, which otherwise I am highly unlikely to attend. And as I mentioned here, and here, it was rather good last year, to say the least.

And thirdly, the Safari theme suits me. If you have to find a dataset to work with and spend a long time on creating a viz, then I would always find something you enjoy or are passionate about. I have long been passionate about wildlife conservation, and so when I was browsing the Environmental Investigation Agency website and found some data about rhino horn seizures, I knew I had the start of what I wanted to to. I was about to enter my first Iron Viz. In fact here’s a photo of me from seven years ago … (and yes, the medal proudly displayed on my horn is a London Marathon completer’s medal)

4168_83883865333_4097324_n

I wanted to challenge myself in several ways. I wanted to emulate some of the best Tableau Public authors around, to challenge myself in the use of more artistic layouts. 100% floating tiles, transparent images to act as backgrounds, use of photo images, long form infographic style, circular insets to name a few things. I wanted to emulate the master of such visualisations, Jonni Walker:

Dashboard 1-107

Jonni’s viz above is one of many he has done on different animal and bird species. It’s so good, as well as looking like it has come straight out of the pages of a magazine, it actually made me want to go out into Derbyshire and look for kingfishers. We talk about a good visualisation having a call to action. There’s no more literal call to action than that!

Fast forward to last Sunday. My visualisation was making progress. I’d got most of what needed including working reasonably but still with plenty to do. And then, a stunning early entry from my good friend Ken Flerlage dropped on Monday. To emphasise, my version below is far from finished! But, see if you can spot some similarities and differences:

 

So here goes. Ken’s is amazing, mine is unfinished. Ken’s is black, mine is beige. Ken’s has a hard-hitting message, mine includes the same issue but doesn’t tell the full story, instead with more generic rhino info. Ken’s has great text annotations mine doesn’t (yet). Ken’s is polished, stunning and deserves to win … mine, well you get the point But we have both gone for exactly the same story and the same issue. We’ve even used exactly the same datasets from exactly the same sources. We’ve quoted and recommended almost exactly the same charities. We’ve both gone for circular insets and long form dashboards (of exactly the same length!). Given the only guide was “Safari”, it’s uncanny how similar our submissions are.

Ken’s blog is here – it includes some shocking imagery but I make no apologies for that given the subject matter:

http://www.kenflerlage.com/2017/06/the-killing-fields-rhino-poaching.html

The image on the right is exactly how it was the moment Ken got his early entry in – I mailed it to him to show just how close our thinking has been, and so it wouldn’t look overly suspicious if I entered at the weekend with such a similar idea. So what do I do now? Anything I do will either seem inferior or look like it’s copying. I know mine is far from finished – I need to decide what to do at the bottom, I need to sort out text and annotations, I need to finalise the story, I need to update tooltips and interactivity. And I’m not overly happy with the distribution map, since the specks for the  Javan, Sumatran and Indian (greater one-horned) versions are barely visible. A lot still to do, but now I know there’s such an obvious yardstick!

It doesn’t feel right to enter now. I will finish the viz – otherwise this is too much of an excuse not to. And I entered because I wanted to try my first ever viz of this type and highlight an important issue. Thankfully Ken is highlighting this better and more powerfully than I could have imagined! But if I do enter, it will seem like I am trying to outdo Ken’s viz, and I don’t think that’s even possible. I’ll think about this. If there are things I do that are slightly different (do I focus more on the map and the information on individual instances, do I include a video, for example?), is it because I’m trying to gain an edge? If I focus on things that Ken has included too, is this just imitation?

Ken (sorry Ken, you’re getting a lot of mentions here!) is one of the nicest men I know in the community and has often been very supportive and appreciative, so I know he will think there is nothing funny going on. In fact, he’s encouraged me to enter, despite us both having no idea we were working on identical data and issues!

Incidentally, a couple of weeks ago, a fascinating blog was published on the subject of tile maps and maps of Africa, using the recently published data of world countries’ Internet usage.

It was Ken again! Ken has done some amazing work considering many more regular and irregular ways of representing Africa using tile maps (the link to his blog is further up this article). Only during the last phase of his work did he realise that someone else was doing similar work and blogging about this; er, me – here and here! But he didn’t mind at all and nor did I.

The coincidences continue, perhaps they’re not *quite* so coincidental if you consider that there are two of us out in the blogosphere who are both fascinated by Africa, its maps and its magnificent endangered inhabitants. There is certainly room for the works of two keen data amateurs to explore different ways of visualising the continent of Africa, so perhaps there is room for two takes on the same important issue, if only because it gives twice as many opportunities to highlight the organisations that are doing great work trying to combat crime and poaching in this area. Ken’s been kind enough to call me out in his blog and I’m delighted to do the same for him.

I know who I want to win iron viz! That said, as always I’m excited to see the amazing standard of entries and looking forward to many more. I just hope there are no more on rhino horn trafficking, that might just be one too many …

 

Why do we visualise using circles?

Why do we visualise using circles?

I should clarify – if the books of Tufte and Few are your visualisation bibles then you may not enjoy this post. Rightly so, the two authors I’ve just mentioned are seen as the foremost proponents of the field of creating professional, analytical dashboard visualisations.

But for those who enjoy data visualisation as data art, who appreciate the aesthetics of data visualisation over and above the rigorous application of best analytical practices, then, like me, you’ll know that the data visualisation field spreads far wider. So I was excited to see the publication of a new book from Manuel Lima, the Book of Circles.

I have done a lot of visualisation using circles – more than I’d realised until I came to contemplating this book and the review, with all of the below featuring in my portfolio, visualising subjects as diverse as football results, 1980s music, Donald Trump’s tweets and animal counts in a zoo inventory.

To summarise this book I was able to read it from cover to cover on a long train journey, which means that (a) it wasn’t too text-heavy that I couldn’t complete in a couple of hours and (b) I was captivated, and once started didn’t want to put it down.  If you don’t want to read further than my overall rating, then my key takeaway is that as a result of reading the book, I wanted to do three things:

  1. Read it all over again
  2. Go away and do some (more) circular visualisations, inspired by those in the book
  3. Recommend this book and tell everyone how good it is.

And I’m doing these three things right away, starting with number 3.

The first thing to note is that this leans more towards a “coffee table” book than a technical reference book. The majority of the book reads like a gallery, with full or half-page high quality visualisations and gallery-style descriptions. But, despite this, strange to say that my favourite page is the page below:

IMG_0282

The contents page “taxonomises” circle visualisations into seven different types, with different sub-types within them. Instantly from this I can reference those that i’m familiar with and those less familiar with, and can see potential ideas for adapting work I’ve done to fit a new circular visualisation type. It means that if you want to derive inspiration from concentric ring visualisations (for example), you know just where to turn. Though I challenge you not to be drawn in to all the other visualisation types in the book in the meantime. I do wonder though, which one would best represent my concentric circular bump chart? Maybe an idea for volume two?!

The first fifty or so pages are just the right amount of theory to discuss the aesthetics and psychological use of circles over the years before launching fully into the visualisations for the remainder of the book. But these introductory pages still contain many more examples of circles through the ages. Particularly relevant is the explanation of the metaphor of the circle in four ways:

  1. Perfection (simplicity, balance, harmony, symmetry etc)
  2. Unity (wholeness, completion, containment etc)
  3. Movement (continual force, rotation, cyclicality, periodicity, etc)
  4. Infinity (eternity, immensity, etc)

Lima’s examples throughout the rest of the book span art, astronomy, biology, cartography and of course the more recent notion of data visualisation. Allowing for coverage of such a wide range of genres, he intersperses ancient, classic and modern examples throughout to show how our concept of circular visualisations has lasted over the centuries. I’ve toyed with the idea of picking some favourite visualisations from within the book to show you but I’m not sure that’s fair on the original artists. Beauty is obviously in the eye of the beholder and if you read this while leaving preconceived ideas of data visualisation rules to one side, you’ll see several hundred eye-catching visualisations.

Instead, I’ll leave this circular image – the one chosen for the book’s cover, and suggest this would make a  fine addition to the shelves of any data art enthusiast

IMG_0284

Do we publish visualisations just for the attention?

Do we publish visualisations just for the attention?

A week or so ago I wrote three-quarters of a blog post about this, inspired by the attention that my Africa tile map viz was getting. I ran out of time and “parked it”, heading off to the Tableau Conference in London

To summarise my first (unpublished) post: I compared a data visualiser publishing her personal work online to a (traditional) artist displaying her work at a gallery. An data visualiser’s collection of work on his blog, website or Tableau Public profile is like an artist’s portfolio. I was pretty sure I had a good analogy going, but that’s what it was, an analogy. And it was an analogy that meant i was prepared to admit that yes, I do publish visualisations for attention, and this contextualised it nicely for me.

The portfolio analogy of the online data visualiser was a simple and obvious one, but a key one that bears repeating. I’ve often felt that a varied and eye-catching portfolio of visualisation work is a huge help when seen as just that – I know of several people whose portfolio, in Tableau Public or otherwise, has actively helped in the process of getting a job (myself included).

But, as I say, I didn’t quite finish that or draw it to a reasonable conclusion. I went to the conference last week, where in the “Box and Whisker Bistro” (one of the many themed rooms in the venue) I saw this:

 

It’s worth saying, that at my age (and those with a keen eye and an accurate abacus can work out exactly what that is from above), things can come as a surprise even when you’ve had strong hints dropped about them in advance – I remembered there may be such a display, but had subsequently totally forgotten. It’s great – I can watch my favourite films over and again and still not know what happens! Anyway, back to the blog …

So what’s this? It’s a real actual art gallery (for the duration of the conference). With actual art works. And they picked mine, not once but twice. Now, I’ve always loved the idea that for more creative visualisations on the David McCandless side of the spectrum (look away, Stephen Few and advocates) would be great if framed, blown up in size to large installations, and shown as artworks, but I’ve never seen it done. And it’s fair to say, I bloody loved it.

And they’ve even put biographies for each picture (shame about the date of birth) to make it look like I’m an actual proper artist. A nice touch!

Here’s a photo I took of Tableau’s Andy Cotgreave (there were whispers that he’s written a book recently that might be very good – the Big Book of Dashboards and Charlie Hutcheson – a great friend in the community who has quite possibly the best blog for those learning Tableau. Not only does Andy have a great current book, but I believe he was instrumental in the gallery idea. How many chances will there be to meet them both at the same time and get great photographic proof? But instead I shoved them off to the side so I could sneak in a photo of my pictures in situ in the gallery with people looking at them …

IMG_0181

So now before I meander onto the rest of the post I should give some context and a big thank you to Tableau. The context: I’m a non-artist. A non-drawer. My artistic talent is off the scale in a minus direction. I came from a school where we had to submit sketches to our art teacher which would be graded 5 points for an A, 3 for a B and 1 for a C. Lowest scores got shamed and put in a league of shame on the wall, with potential detention for those not meeting a certain threshold. (Different times!). My sketches got ticks. An acknowledgement that they were done and submitted, but saved from the embarrassment of a mark lower than C. Not saved from the embarrassment of a place in the hall of shame on the wall though. So that’s some context having been through school last century as “boy least likely ever to get work shown in an art gallery in his life”. So it’s in that context that I sincerely thank Tableau for doing this – it made this middle-aged artistic klutz taking up a new career in data visualisation / data art very happy and proud of his achievements – an unexpected career highlight for which I thank you sincerely!

My pieces were just two out of eleven and I’m honoured to be in the company of the remaining works, see them here:

The remaining artists are Rody Zakovich, Chris Love, Matt Francis, Pooja Gandhi, Rob Radburn, Adam McCann, Mike Cisneros, Ben Davis, Many of whom I have previously cited as my influences, and the rest of whom I look up just as much. Thanks to the biographies, I know there’s a certain irony in the fact that I always look to Rody for inspiration (code for I quite often steal his work, like an artist of course), yet he is a depressing eighteen years younger than me!

So, apart from this being a wonderful ego-boosting surprise example of an analogy taking form in reality, have we answered the original question? My thoughts about publishing data visualisations for attention were obviously more leaning towards publishing (and publicising) online visualisations for all to see – whether on social media, blog posts or areas such as Tableau Public.

There’s a feeling that people are using things like Makeover Monday to make a name for themselves. Or that people are courting attention from new or highly technical visualisation types. Those of us who are not professional data journalists or visualisers (perhaps, like me, they will use tools such as Tableau professionally, but not to produce the kind of work that they will publish publicly) will produce a visualisation, but what then? Do we just keep it to ourselves, or do we publish it and publicise it with a tweet or a posting?

I agree, that’s what people do. I know this, because I do it. For me, Makeover Monday is a great yardstick. Last week, I published my Africa tile map visualisation. And, with no great subtlety, because it’s one I’m really pleased with, I’m going to show it again here.

Dashboard 2-16

Because work published for MakeoverMonday gets published at roughly the same time each week, to the same group of Twitter followers, I know that certain works are getting more attention than others. And whether we court it or not, (positive) attention is the sign of a visualisation that has worked well. I make no apologies for excitedly watching and counting the number of likes and retweets it garnered through the week, because I could draw an obvious conclusion from it: that was a good one, people liked it.

Week 22: Internet Usage in Africa: 125 likes, 28 retweets

Week 21: Drinking Habits of Britons: 12 likes, 0 retweets

Week 20: Youth Employment in Latin America: 27 likes, 0 retweets

Week 19: Car Colours in the Netherlands (seen above, which I’m not sure if I’ve even included in this blog before, and therefore shown below): 44 likes, 3 retweets

Week 18: Sydney Ferries: 23 likes, 1 retweet

Dashboard 1-105

And so on … whereas it’s not a matter of pimping likes and retweets, the reaction of a visualisation in the public arena is as good an indication as any that we have of how well received our work is. I can look at the above, and see that my week 21 submission was largely overlooked. OK – I have to squint through the numbers and it would be much clearer if I visualised it, but you get the point. Week 20 was also somewhat down on my usual figures. I don’t see it as being competitive, but rather see the quantity of interactions as a critique on my work. Weeks 20/21 , could do better. Week 22, you nailed it – you did something good, or at the very least did something that got liked and noticed.

I don’t see it as desperately looking out for likes/retweets and attention from a numbers point of view. I don’t expect anything to “go viral” – I’ve never put anything on reddit and wouldn’t want to. I follow my reactions enough to know that 125 likes for Africa tile maps is my best ever, but I’m fully aware that in the grand scheme of things that’s not exactly a sensation, nor is it likely to get me included in the next Olympics’ opening ceremony. My cousin’s dog, Sprocket, got national media coverage and 3000 likes in a day, and rightly so. He’s the standout picture in this blog post!

sprocket

The “work/out of work” dilemma is a difficult one – many people use visualisations and hone their skills in a certain direction while using software at work, but often to a corporate standard or a very closed skillset. Publishing “non work” visualisations can take any form – do you hone these skills further to create corporate style dashboards on a variety of open datasets, or utilise your hidden graphic skills to create a work of beauty? But for anyone who does so, who is their audience?

Chris Love wrote here (and he’s not the only one) about the difficulties of finding an audience for personal projects. A new, ground-breaking or artistic piece of work will tend to be published only to your professional or social media bubble. We create visualisations considering the number one rule “know your audience” but for non-client dashboards you can’t determine your audience. I could create a piece all about golf, or pop music, or the Oscars (all of which I have done) but my audience, by definition is not specific to any of those areas. Rather it is those who have chosen to follow me on social media outlets, for whatever reason that might be. So it’s easy to see the frustration, or the futility even, of just releasing work to an undefined audience, with the only measure of reaction or appreciation being in the likes/retweets or equivalent reaction. So sometimes, unless we have a different “way in”, that’s our only option for “non work” pieces. But I don’t think that’s something to shy away from, or deny that it’s what we do.

At the conference, I was asked to give an interview for the Tableau Wannabe podcast. I was delighted to say yes (after all, this is a post loosely related to the humblebrag and pimping for self-promotion, so in for a penny …!) just to give my perspective, and I found to my surprise, that as someone who doesn’t usually say much, that give me a data visualisation topic and I can waffle for England!

I do believe though that if data visualisation is both your hobby and your profession, then the public forum is the ideal outlet for what you create. One of the things that I mentioned, which I have had the most feedback on since doing the podcast last week, was about working on personal projects and working with data you love. It’s already inspired some work such as this by Sarah Bartlett (or, if it hasn’t, and they were going to do it anyway, they’ve been kind enough to credit me). And I think it’s important to work on personal projects, enjoy yourself, learn your skills as you’re doing them, and get them out there! If that counts as visualising for attention, by soliciting reaction, feedback and “likes” then count me in.

Perhaps because I still think of myself as being very much on the steep, early, upward phase of my learning curve, I see no shame in this. We’re lucky to have a thriving online data visualisation community who help fellow participants to develop. And if any of those and my fellow friends in the community are reading this blog post and enjoying it, then I ask ashamedly, feel free to like and retweet the link!

 

 

Which shapes work well with tile maps?

Which shapes work well with tile maps?

Last year I posted about tile maps and the fun I had devising one for Africa in particular. My approach up to that point was simple – just devising a square grid, basically based on graph paper and an atlas. Or at least the slightly more up to date version of rows/columns in Excel and an online map.

On that occasion I tried Africa and France, and have since published square tile versions for Scotland and New Zealand using a similar low-tech technique. Last month I tried something a little more complicated, but only a little more, by trying a hexagonal tile map. My choice was Europe (well, Europe and Australia, but then that’s the Eurovision Song Contest for you) with the resulting tile map below:

Dashboard 1-95

Logistics about setting up a tile map notwithstanding, Eurovision is an ideal subject for a tile map. The reason for this is that each country is equally as important as any other, with exactly the same weight given to a vote from (or a performance by) a small country as for a large country. But the first issue with tile maps will always be the disparity in size between countries. There’s no getting away from the fact that in the European example above, Russia and San Marino are the same size, despite the fact that you can fit over a quarter of a million San Marinos in the area of Russia.

The second issue is often the fact that you might not have data for every country/region, whether due to incomplete data, disputes/changes in territory, or small city/state areas which are often not included. My Europe example has many of those issues. I’ve included Andorra and San Marino because they have been included in the contest, but not Liechtenstein, since they have never taken part. I have included Luxembourg, since they have participated in the past, but they have ended up as a blank tile. But not included is Kosovo, only recently recognised as an independent state. This is more common than you might think – does your America data include D.C? Does your UK data consider Bristol as a separate county or part of Gloucestershire? Is Kosovo recognised as a separate state? Is Greenland a separate entity in your data? … etc

The third issue is just one of aesthetics. Does it work? Tile maps aren’t for everyone, and none can be done without a certain amount of displacement and distortion. But is the overall shape/format recognisable as the country or continent it represents? Plainly and simply, does it look nice?

To compile a hexagonal tile map (in my case, in Tableau) the method is very similar. You can even download hexagonal grids if you want to print off and colour your map while planning manually. The only difference is that as a “grid”, some of the columns will be offset by 0.5. This is really well explained in this post by Matt Chambers, aka Sir Vizalot, and so I used the techniques he explains to generate my hex tile map in Tableau.

Last month in Manchester I was offered the chance to speak at a Data Science conference (as alluded to in my last post). I decided to attend the full two days of the conference, partly to acclimatise for my first major conference talk, and partly because it seemed like a great conference, which the organisers had made me feel very welcome to be part of. Scanning through the list of talks, I noticed that Dr. Graham McNeill from the Oxford Internet Institute was talking about how to automatically generate tile maps. He and his colleague Scott Hale are to present a paper on their findings at Eurovis in Barcelona later this month (June 2017). In a conference where a lot of high level mathematics and science was included (my talk, on the second day, was the first not to include equations, algorithms or Greek letters!), this stood out to me as one that really reached my interest.

Graham’s talk explained that he was devising a program/application that would convert input locations to a given tile map, using algorithms where users could determine the most important features. What is most important – overall shape? Positioning? Geographical accuracy? Are there certain arrangements that can’t be broken? For example, he showed several different interpretations of US tile maps, probably the most widely recognised tile maps, explaining that different publications use slightly different “house” versions. Some have the most optimised versions with North and South Carolina actually adjoining east and west, whereas some have North above South, and need to make concessions in other areas. Problems I’m well aware of from my low-tech map generating!

The other thing that attracted my interest was the fact that he was easily able to toggle between four types of polygon: square, circle (essentially the same thing – the arrangement is identical, it just doesn’t tile snugly), hexagon and triangle.

(Wondering how exactly a circle tile map could work? Take a look at this from https://flowingdata.com/2017/05/18/chernoff-emoji/ with further explanation within)

Emoji-states-of-America

The first three (square, circle, hexagon) I’d seen, and tried, but not a triangle tile map. The talk was therefore exciting on two levels: firstly, is there a method in the pipeline that will do the mapping part for me, better than I can do myself (though secretly that’s a bit of a shame, as it’s the best bit!), and secondly, regardless of the algorithm or not, it completely opened my eyes to the fact that a triangle tile map was an option. We were shown a few examples, some of which looked better than others, but it was quite exciting to think that a triangle map could look good in certain cases.

Graham mentioned that he’d tried two in particular, because he’d seen some good manual versions of France and Africa out online. Cue moment of surprise and the slide below (for those who haven’t followed my blog or clicked on earlier links, these are, of course, mine!). As a newcomer in the field of data visualisation, I was quietly chuffed to see my work (via this very blog) had been noticed and publicised.

IMG_0142

So this got me thinking – could we work together to see whether there were some interesting tile maps that would work for data visualisations that I could try using in Tableau? Shortly after the conference I asked Graham if he had some interesting triangular maps to send me that his application had generated.

There were three versions of Africa:

 

The first thing to notice about triangle maps is that there are two different types: essentially those where the triangles are all on their point/base (so the bases align horizontally), and those where the bases align vertically. The one thing that a human has over an algorithm is their own aesthetic choice. I don’t know which of the three is most accurate, but instantly I saw one of them (the largest of the three images above) and loved it. I prefer the former kind to the latter kind, but that might be because of the longitudinal nature of Africa – perhaps a different-shaped country or continent would suit the alternative alignment better?

The algorithm had done a great job of aligning Africa into a really pleasing shape, but there were three things that didn’t quite sit right. Morocco pointing up out of the horizontal plane in the North, Somaliland poking out to the East, and Lesotho occupying the furthest tile south. Along with the image, Graham supplied me with a row and column number to use to import into Tableau, and one extra parameter, essentially whether the triangle is to point up or down. This is the only difference to a square or hexagonal map, since two separate alignments are needed within the same map in order to tesselate.

Fortunately this is where overriding some of the choices manually becomes an option, and some of the quirks of African nations worked nicely in my favour. First, I swapped Lesotho and South Africa, so that Lesotho sat above South Africa instead of at the very bottom. Perhaps this is the equivalent to the North/South Carolina quandary I mentioned above, but it didn’t sit right without having South Africa at the bottom, especially since Lesotho, though entirely surrounded by South Africa, is not a coastal nation. Secondly, Graham’s map included Western Sahara, which is still considered a disputed territory. The dataset I planned to use was the latest Makeover Monday (you may have heard me mention that before) which didn’t have Western Sahara, so I was able to manually adjust the surrounding countries and smooth out the Northern coast. Thirdly, Somaliland was also not included in my data, and a quick google suggested it’s also not a recognised country.

Finally, the island nations (Sao Tome & Principe, Cape Verde Islands and Madagascar)  weren’t included in the original algorithm. Those just need to be added manually.

So I had my version of a nice smooth tile map and used small multiples to show each of nine separate years. My design choice was to use ochre shades of brown/orange, but much of the rest of my design choices were down to a bit of luck. I struggled to get the triangles to line up, and decided it didn’t matter too much. The added zig-zag lines, as well as leaving a bit of visibility between nations, alluded to African patterns, and really helped the design theme. The resulting image, below, was one I was very happy with:

Dashboard 2-16

The feedback has been quite overwhelming – it seems many of you are appreciative of a new type of tile map not seen before. In particular, Tim Ngenwa in his feedback pointed out the following page https://uk.pinterest.com/sharonlwhittake/african-patterns/ which included African patterns that it resembled, such as the pattern below:

c85214e1530c39026a64bb91e708474e

From a Tableau point of view, if anyone was to download my workbook, you’ll see how simple it is. The only slight difference to the work I’ve done previously is that the row/column values are not 1,2,3 (or 1.5, 2.5 etc), but numbers relating to the latitude/longitude of countries (and to several decimal places). But once brought into Tableau, because the points will be shown to scale, and the axes suppressed, it doesn’t make any difference.

Not all tile maps work, and there will be tile maps that work well with certain tiles, but not so well with triangles. With the UK election on the horizon, I’ve been badgering Graham for a UK constituency triangle tile map, but the upshot is, with the algorithm as it stands at the moment, it just doesn’t look very good! The non-smooth geography of the UK makes it a difficult proposition. And 650 constituencies probably make it a leap too far for a manual approach (though I won’t rule it out!) In addition, I’m not a big fan of the US tile map above, whereas the established hex and square tile maps do a good job. But Graham’s France map (below) looks promising! So I’m looking forward to investigating more tile map options.

france_base

So expect more tile maps from me in due course – all feedback welcome, I may even consider requests!

Should data visualisations always tell a story?

Should data visualisations always tell a story?

I feel a bit of a fraud asking this question. It’s been asked, debated, blogged and further debated many times before. You can get opinions on the subject from people with much more experience in data visualisation than me. For example: Cole Nussbaumer Knaflic, Alberto Cairo, Jon Schwabish, Georgia LupiChad Skelton, Sophie Sparkes, Robert Kosara to name but seven of many. So do please check out what they have to say, but allow me to put my thoughts here too, even if that just involves collating and arranging the thoughts of the aforementioned others.

So why am I even broaching the subject? Two reasons really. One – I daftly decided to start a blog (you’re reading it) about questions in data visualisation, and what better question? And two – in less than a fortnight I am giving a presentation on data storytelling at a conference in Manchester. I’ve volunteered for this, stepping into the breach to help out a presenter who couldn’t attend, and have started off with obvious fear at what would be my biggest speaking engagement so far. The more I’ve researched, read and listened to many talks on the subject, the more excited I’ve got at the prospect at having a great subject to talk about with so much potential material – I should have no problem talking for 45 minutes about a subject that fascinates me within data visualisation. And yet the more I’ve continued the process, the more daunted I’ve got again. There isn’t overall agreement, there’s no real answer. How can I tell a hall full of delegates what data storytelling involves when there isn’t agreement within the industry?

So here’s a frivolous but pertinent example from none other than Kurt Vonnegut. Known for his modern science fiction classics such as Slaughterhouse Five, Vonnegut also showed how stories can be plotted out on a simple pair of axes:

Ill Fortune to Good Fortune, Beginning to End. In the story of “Man in a Hole” (which requires neither a man nor a hole) the protagonist starts off fine, then finds some trouble, and gets out of it in the end.

In Vonnegut’s diagram of the “Boy Gets Girl” story, the boy finds something wonderful, then something bad happens (e.g., the girl and boy fight), but then he gets it back again.

VonnegutStructure-01

Two examples of how the simplest and most classic of storylines can be visualised graphically – as (almost) a sine curve. I could flip this upside-down and change the words (if I had the inclination and Illustrator skills), then you’ve got a very true to life story:

  • Middle aged Man prepares conference talk: by Neil Richards
  • Man agrees to talk – is worried and daunted (ill fortune)
  • Gets lots of information, gets really interested (good fortune)
  • Keeps researching, realises there’s an overload of contrasting material and gets daunted again (ill fortune)
  • End

So where to start? Last week, Cole Knaflic, author of Storytelling with Data visited London to talk to the Tableau User Group. I was privileged enough to attend. As you’d expect, Cole explained how visualisations could be improved by treating them as stories with all the elements you’d expect in a work of fiction. Your work of fiction doesn’t have to be Pride and Prejudice, it can be an adventure book written for a small child. But all good fiction stories have characters, a theme, a plot, a point of view, setting, conflict and tone. All these were great pointers to improve our presentations and business dashboards. The result of this approach was to highlight the importance of many data visualisation principles.

In the two hour train journey to London I finalised this visualisation of the French Election first round results.

Sheet 2

Cole mentioned that colour is great – for children’s parties. If you don’t need it, don’t include it. An hour into my journey home, I had the following. It still needed some colour, after all there are four main candidate, but this was far less busy and made every outcome clearer. Much better – an instant improvement, and a story more clearly told.

Dashboard 1-90

I asked Cole the same question as the title of this entry – should every visualisation tell a story? Her answer was “no”. There’s certainly a distinction to me made between explanatory and exploratory visualisations (a distinction I’ve attempted in the past myself). But the key thing is that a visualisation should always convey the message that it’s intended to convey.  Her blog then makes a distinction between a “Story” (with upper case S) and a story (sic). I’ll leave you to explore the differences, but perhaps the conclusion is that every visualisation should tell a “story” (lower case s!)

Jon Schwabish is keen to point out how the term “story” is overused, that many visualisations in fact do not tell stories. I’m not sure whether or not he concludes they *should* but don’t, or just don’t! But I do like Chad Skelton’s interpretation a lot. He takes us away from the idea of a story in terms of a work of fiction.

Fundamentally, the purpose of literary stories is to entertain and the teller of a fictional story has the luxury of making things up to ensure their story is as entertaining as possible.

In contrast, the purpose of data analysis is usually to inform an audience and those visualizing data are limited to plain old facts.

What if those facts don’t stir emotions, or fit into a satisfying story? Trying to make facts “tell a story” could distract from communicating correct analysis. Chad suggests that a “news story” is a far better alternative. Inform rather than entertain – you can’t make something more entertaining than it is!

However, I do think it’s possible to tell an intriguing story (OK, a news story) by careful selection of data. The next examples are league tables – they are not sophisticated or artistic, but they are every bit as valid as an example of data visualisation as a bar, pie, bubble or any other chart you might want to mention. If I asked who was the best football team in England, would you remember the most recent complete season and give the answer as the runaway league winners, Leicester City?

Screen Shot 2017-05-04 at 23.19.10

Or would you say this year’s league leaders and favourites for the title, Chelsea, who have as many points as Leicester did last year, with four games still to go?

Screen Shot 2017-05-04 at 23.19.47

Both are credible suggestions, but if you were to consider the overall results over the last two seasons, a new team emerges – despite not topping an individual year’s league table since the 1960s, Tottenham are a clear 15 points ahead (graphic courtesy of the Sun).

Screen Shot 2017-05-04 at 23.22.25

This isn’t so much a data visualisation technique, but it is a data storytelling technique. And hard to argue, when facing the less conventional third option above, that Tottenham are not the best and most consistent team in England at the moment.

Whatever your definition of a story, it’s possible to make your data tell the wrong story. If you’ve seen this chart before, featured and debunked in Business Insider, you’ll know exactly what’s wrong with it. If you haven’t – I’ll make it easier and put two at first seemingly different charts next to each other …

There are lots of powerful (and, unlike the above, ethical) ways to tell a good story with your data. Whether that’s with a small or large S, these ways give extra impact to your data. And that’s what I’l focus on when I talk about storytelling, not so much about the semantics of the word, which will continue to fascinate. I’ll probably include the above charts and will definitely include very many more, and my aim is twofold: to promote interest in great data visualisation, and to stimulate debate rather than answer questions.

Still terrified though!

 

Why have I stopped contributing to Makeover Monday?

Why have I stopped contributing to Makeover Monday?

It’s difficult to be a regular contributor of Tableau data visualisations online via Tableau Public without knowing about the excellent Makeover Monday initiative. For those who aren’t aware of it, I penned an ode to it last year. As last year continued, I posted a contribution every week, from the simple to the complex, learning something new each week. Towards the year’s end, I even decided to go back to the start of the year and complete those from weeks before I started.

Makeover Monday has been good to me – not just a way to test my skills and learn something new each week, but it’s really raised my profile, to the extent of recognition from Tableau itself. Here’s an unsolicited T-shirt by way of a gift from them which I often wear proudly with my name emblazoned on the back (OK, not usually in public, but still …)

IMG_0022.png

Since then, it has gone from strength to strength. This year it is being run by Eva Murray and Andy Kriebel, with, from memory, about 140 unique contributors every week, from all around the globe. In a new move, they will report back on their favourite 4-6 contributions of the week, with great feedback for all the participants of all the good and bad things seen through the week.

So, wait, doesn’t this post say that I’m *no longer* contributing to Makeover Monday? Fifteen weeks in, and I have contributed fifteen more visualisations, proudly sitting with sixty-seven out of sixty-seven weeks completed.

Let me explain.

I’m sure the reboot at the start of the year resulted in a lot of new goals. New resolutions, from people wanting to get involved more regularly. I actually made a resolution as well, but resolved to do the opposite: my previous year had been so good that this year I wouldn’t feel I have to do everything that came up. I’d loved Makeover Monday, but I’d been involved every week, and no longer had to in 2017. I can pick and choose now, choose other projects and ways to advance my skills and experience. Maybe I don’t need Makeover Monday any more. I’ll still take part from time to time if it looks interesting, and if I have enough time to spare on a given week. The weekly roundup of favourites still seems like a good goal/aspiration to have, but I can attain that without entering a visualisation every week. But those will be my terms for the year. And that’s why I decided to stop contributing this year.

So 2017 began with a makeover of some data published by the Australian government about the gender pay gap. I was happy to do my version of the makeover, but it occurred amid quite a controversial week. The reporting method of the original data source was questioned, and there were strong feelings in the community about whether or not it was the responsibility of participants to investigate and call this out. Though the argument was valid, it didn’t feel comfortable, nor did it feel like the right time to leave the project. Before I knew it, it was time for week 2, and a less controversial makeover about Apple sales figures was upon us. I could stop my regular weekly contributions on a happier note now.

Week 3 of my “non-contributing”, and a fascinating dataset about the tweets of Donald Trump. This looked a fun set to get stuck into and teach myself something new, so I didn’t mind contributing to this one. In fact I definitely didn’t want to miss this – the result was my spiral visualisation below, which I blogged about here (click the image for interactive version).

dashboard-1-66

A lot of appreciation for my Trump visualisation. No call-out in the Makeover Monday favourites for the week, but no matter. It’s unorthodox and faced some fantastic stiff competition. Week 4 – wait a minute, I can invent a new tile map for this one! Having had such fun and success working on my Africa tile map last year, this would be a good opportunity to try a New Zealand tile map. The result (and the improvements made on it by others in the community) was blogged about here:

Weeks 3 and 4 had got me back into the swing of things, and before I knew it weeks 5 through 8 were completed with visualisations on employment, Chicago taxis, Valentine’s Day and potatoes. Nothing groundbreaking from me, but the fantastic contributions from the community meant I wasn’t really ready to step away yet. Weeks 9 and 10, on credit card spending and YouTube channels started to make me revisit my intentions at the beginning of the year, as I spent more time practicing, polishing and visualising my contributions but not learning a huge amount more in what i was doing. Most of what I learn now comes from the jaw-dropping standard of fellow participants rather than anything new from myself.

Finally, week 11 was officially going to be the week I stopped contributing. A risqué visualisation on the frequency of orgasms. I don’t like small datasets, this had just 6 numbers. The results – well, let’s just say they presented no great surprise to any man or woman, so there were no analytical nuggets of wisdom to be found. I don’t feel I have infographic/creative skills, and as for the subject matter … well, we’re all grown up consenting adults (apart from poor 8-year-old Joe!) but it’s not really a subject I envisaged visualising. I sat this one out in the knowledge I’d made the break at last, and waited for the weekly round-up.

Finally as a non-contributor I had my freedom! From now on, as I resolved 11 weeks previously, I’ll just do the ones I feel like. I don’t need another Tableau T-shirt, or the commitment of a couple of hours a week for a new viz (they don’t take just an hour for most people, I don’t care what the guidelines say!)

Week 12 – March Madness. A huge detailed dataset, about sport. Perfect for me. I don’t mind that I’m no longer at 100% now, I’ll do this one. It’s a perfect opportunity to try a chart type I’ve been planning, adapting a previous viz of the day from user Shivaraj. At this point, I realised it would take me a fair amount of time to set up and execute, so … I fell back on the wagon. Yes, I quickly did a late contribution for week 11 – it would be a shame to drop from the 100% completion rate for the sake of a simple slope graph. After all, it was only 6 data points to visualise!

A big thank you to Shivaraj for the inspiration, who himself drew it from a detailed blog post from Tableau guru Bora Beran – another great example of the openness of the data visualisation community and Tableau community in particular. Below is my March Madness visualisation, showing the progress of different seeds in the first and last year of competition (1985 and 2016).

march madness-2.png

A lot of recognition from those who enjoyed this viz, and probably a lot of shrugged shoulders from those who didn’t! The latter group probably included the Makeover Monday organisers, and that’s completely understandable, but I’m really happy with how it turned out (with apologies to those who thought that the round circular skin tones of the visualisation meant it resembled something a little risker, sorry about that! I suppose it’s a downside of doing so many circular visualisations!).

Free from the constraints of contributing every week, I fancied a go at week 13 – the “secrets of success” was an ugly representation of some questionable Russian survey data. I was pleased with my makeover but the community was non-plussed, however there was a lot more debate on how, or indeed whether, we should visualise data which at best left as many questions as it answered. I’m strongly of the opinion that Makeover Monday is a visualisation exercise only – if the data source is correctly attributed then it’s not up to us to critique and analyse the research, only to display it accurately and without errors. When the commentary around each of the projects focuses more around issues of data collection, research validity and shouting down the original data report (as opposed to the visualisation) then I might feel it’s a valid argument, but it’s a shame, and my interest fades slightly because that’s not what I want to be involved in. And so, by week 14 I re-affirmed my intentions of week 11, and had no intention of competing unless I could find an angle of learning or something to pique my interest.

Well, I found it – it was Marimekko week (for me, anyway). Here’s the visualisation I created but the main blog post on this can be found here:

Marimekko-2

Now, I try not to be competitive, and really I gain so much personal satisfaction from those visualisations I feel have gone well, and take lessons from those who do so much better than I do, particularly on the weeks where my entries are less inspiring. But I really fancied a shout-out in the weekly round-up here. Personal congratulations from three Zen masters and 40-odd likes for my tweeted contribution must count for something? No, still, nothing!

Now, I should add – I completely and utterly don’t mind! Yes, it’s a goal of mine to be included in a weekly round-up, and I’ll be disappointed if this doesn’t happen at least once this year. To see why my Marimekko didn’t get included, see the quality of competition I’m up against by reading the round-up from week 14  I’m biased, and might have included my contribution if it were up to me, but you can see the competition is phenomenal. But perhaps my goal to be included in the round-up at least once is why I’ve found it harder to let go than I thought! After all, the round-up with Andy or Eva’s expert commentary and advice, and the additional community feedback are newer, even better reasons to get involved in the project every week.

Deciding not to do this every week has increased my enjoyment and given me a little more freedom with my visualisation projects. Ironically the result is that I’ve still continued to produce something each week. I even contributed to week 15 by mistake – I didn’t particularly relate to the dataset or produce anything new, but before I knew it, I’d knuckled down, produced a nice analytical chart (nowhere near the standard of some of the visually stunning best) and enjoyed the contributions of the rest of the community.

Now, I don’t claim that I will do all 52 in 2017, nor am I angling for another T-shirt. But I can’t say enough about the benefits of the project, and, contribute this week or not, I’m sure I’ll be checking my twitter stream this Sunday afternoon secretly hoping it’s a nice interesting dataset. Perhaps the title of this post is wrong, and should read as follows:

Why do I continue to contribute to Makeover Monday every week?

 

What is a Marimekko chart (and when should you use one?!)

What is a Marimekko chart (and when should you use one?!)

If you’ve read previous blog posts, you’ll know that I’m a fan of different and unusual chart types. If the chart types are visually interesting and striking, but don’t necessarily follow visualisation best practices, then, for me, sometimes, so much the better! I’ve blogged about, and attempted, sunburst charts, radial bump charts, bump charts, spiral charts, chord charts, spark bar charts … you get the message.

But when I recently found out about the Marimekko chart a few months ago, even I wasn’t sure if I liked it. I couldn’t really think of a use of such a chart other than just for the sake of it, and though that doesn’t always stop me, I wanted to be sure I could make sense of a use case for it.

So what is a Marimekko chart? Below is a simple example from think-cell.com, the first example you might come across using a well-known search engine. Essentially, it is a stacked bar-chart which totals to a whole both horizontally and vertically. To this extent, each vertical bar has a different width proportional to the overall sub-sample.

sample_mekko

But this week’s MakeoverMonday dataset got me thinking. The original chart requiring a makeover was the chart below, from the Guardian.

Screen Shot 2017-04-03 at 23.06.30

The chart aims to show the chance of each profession becoming automated within 15 years or so, along with the percentage each profession consists of in the workforce. The main issue, for me, with the original visualisation, was that showing two percentages on the same axis doesn’t really make sense. We’re being asked to compare two percentages which aren’t relevant to each other, so don’t belong on the same scale.

My first thought was a scatterplot – plotting percentage of workforce versus percentage at risk would allow us to cluster together professions, or identify traditional quadrants: which professions are both more likely to be automated and have a higher proportion of the overall UK workforce. Something like this (excuse formatting):

scatter

But I wanted a different way to combine likelihood of automation with the overall share of UK workforce for each profession. From a numerical point of view, multiplying the two together will give the overall percentage of the entire UK workforce. And from a visualisation point of view, this can be represented by multiplying height and width together to give area.

Enter the Marimekko chart. With x-axis ranging from 0% to 100% and y-axis also ranging from 0 to 100%, the entire chart area therefore represents 100% of the UK workforce. Realising this as I experimented with visualising the data was the lightbulb moment for me where the Merimekko was concerned. The total area of the bars therefore nicely sums up to the overall percentage of the population at risk, in this case a round figure of 30%.

Marimekko-2

The influence for creating the chart was from Emma Whyte, who herself drew from the great work done by Jonathan Drummey. Emma blogged about this recently as a recent #WorkoutWednesday challenge in this blog post here. I’ve chosen to use the same technique as her, blending all the non-selected elements to the same shade of grey to look like background. The chart Emma (re-)created was this one below (not hard to see where I got the inspiration):

Marimekko Makeover

Thank you also to the suggestion of Steve Wexler – he has a simper version of the same process here, which will be familiar to any users of his excellent and instructive stacked bar charting method for Likert questions in survey data. Be sure to check it out for his thoughts on the Merimekko (initially sceptical and similar to mine). But where it really comes into its own is in visualising two interconnected percentages, both of which sum to 100%, so the total chart area can represent the sample whole.