I was really thrilled to have Andy Kirk on the podcast this week. Andy is an independent expert in data visualisation who I’ve been following for some time and who has had huge influence on data viz design and the mentality with which many people approach it now. Because what we’re talking about is a visual format, I’ve uploaded the whole episode here as a video as well. Watch above and you’ll also see the visualisations that we’re talking about in the recording.
Subscribe/ follow on Spotify or Apple Podcasts and follow me on here for new episode updates.
Some links to things we talked about in the episode:
Andy’s website: www.visualisingdata.com
Inspiring data viz examples:
https://www.reuters.com/graphics/
https://www.theguardian.com/technology/data-visualisation
https://www.dear-data.com/theproject
(Mostly) free dataviz creation tools:
Andy’s recommendation of one to watch:
https://www.not-ship.com/author/amanda/
The original Seinfeld dataviz project:
https://visualisingdata.com/2020/09/new-project-the-seinfeld-chronicles/
TRANSCRIPT.
[As always, ignore the Americanisms and please excuse AI’s transcription errors which I know are in there]
GEORGINA:
My guest this week is Andy Kirk, an independent expert in data visualization. We’ll get onto exactly what data visualization means in just a bit. Andy is a prolific producer of resources on the how-tos of data viz including the incredibly useful and beautifully crafted book, Data Visualization, a handbook for data-driven design. Anyone who’s even dipped a toe into the world of data viz online will have come across Andy and his amazing website, visualizingdata.com. Andy has lectured and taught on academic programs, trained people all around the world, and consulted for a whole range of organizations on how to get visual insight on their data. I’m not sure how he has time for it but he also hosts the very successful podcast and video series, Explore Explain. It’s no exaggeration to call him the king of data viz, although he probably won’t thank me for that nickname. My other personal favorite bit of information about Andy is that he once created an entire book which went into a deep analysis of the TV show Seinfeld. So I might have to ask about that later on. Andy, it’s great to have you on the show.
ANDY:
Thank you, Georgie, and it’s lovely to meet you after many years admiring your work as well. And thank you for that very kind, far too kind bio as well.
GEORGINA:
No problem at all. It’s great to have you here. So briefly to start off, what is data visualization? Are we talking about something beyond just a plain chart, like a bar chart that you might see in a newspaper, or does that count as well?
ANDY:
Yeah, I think it does count as well, but it is increasingly a sort of further reaching, more far reaching field where the primary purpose, the reason for any of this existing is to facilitate understanding, whether that’s you doing your own analysis of data to get a sense of the pattern, the relationships, the findings in your data set, and therefore there isn’t a separate audience, or if there is and you are communicating to others and explaining things, or even just giving them the means to find and discover things themselves. I guess what separates it from any other kind of artistic or just purely graphic design endeavor is that there is the purpose of understanding, is the purpose of communicating kind of intrinsically in this. So whether that is a simple bar charts, a simple, let’s say a traffic light on a street is a visualization, the weather forecast, the weather maps, sat navs, right the way through to some of the more kind of, kind of, things that we see in the media these days. It’s all part of the same world.
GEORGINA:
I love that. Yeah, that’s not something that I’d ever really thought about. But yeah, any kind of representation of sort of signals even could be called a data visualization.
ANDY:
It doesn’t need to be human constructed you know we exist in a in a world where we’ve got some visual cues about danger about opportunities about where to go where not to go the condition of things like leafs the color of leaves gives us indication of what’s going on so all these things are kind of part and parcel of the environment that we live in and we’ve just nicked those bits and put them into kind of created charts and visuals
GEORGINA:
I want to get into that a little bit later on about what exactly is so powerful about data visualization. But firstly, could you tell me a bit about how you got into this field?
ANDY:
yeah i mean i’ll keep it short possible i mean as a kid i was definitely into perhaps two of the main components of math and art at school so i was looking for a kind of field to focus on that would bring so those two will together and but i had to wait until the age of thirty to discover data visit as a as a field of practice and the community practice and the
The situation at the time was I was working with Excel and PowerPoint and doing visual displays of data as part of my job. I didn’t realize it was a thing that had a name and had a label. And by chance, I just did a search for some help with a chart type, the radar chart specifically, and found on the web a white paper by a guy called Stephen Few, who was a prominent kind of author at the time. And it was about the pros and cons of the chart method.
I’d never thought about any sense of discourse about the merits of a different jobs or not before that it was always on a aesthetic level to i like the look of it and it just spot this passion i almost instant light bulb moment of alright this is a thing this brings together to my natural sensibilities and i’m gonna try and get into this more because it sounds like a very cool thing related to my day job at the time but something that i might want to focus on in the future. So that was 2007 and 19 years later still excited about charts.
GEORGINA:
Mm-hmm, still going strong. On the data viz, so you mentioned the radar chart paper that you found. Was it that or was there another kind of, was there a piece of data visualization that you saw that really impressed you back in those days that it kind of indicated you to go down this path as well?
ANDY:
At the time, I was very much into some of the work from the team at the Guardian. So the Guardian graphics team that was led in those days by Michael Robinson, who was really a really important pioneer in this subject, certainly in a public realm. But I couldn’t necessarily pinpoint a single graphic from their collection of great work. Suffice to say that it was just different. was…
bubbles and it was unusual chart, was colorful, it was kind of engaging on that level. But one of the first charts I remember really for the first time deeply thinking about was a string graph, not a very common chart type. But for those who have never heard of it or seen it, imagine a series of sort of wavy lines that kind of form almost kind of contours of a landscape.
And this was a stream graph by the New York Times called the ebb and flow of box office takings. And it was basically showing you over time when movies were released, how much did they take week on week in a sort of rolling average? And then when did they sort of peter out and then disappear? And when did the new ones come in? So this kind of constantly evolving flow of new movies and takings. And it was so different to anything that I’d seen before. But I remember at the time it was a very small field of
people blogging and not even social media back then really. it was the first time I’d seen real deep disagreements about the merits, not just is it good or not, but I love it or I hate it. It was that of polarised passion. And it forced me to, again, just for the sake of learning, what did I think about it? Because rather than just absorbing and following the crowd, I wanted to kind of form my own convictions about whether I liked it or not, but also why.
and I think that’s the key thing, you have to back it up with why. Why do you like something, why do you not? And what would you do differently? And so again, I won’t go into the sort of 3,000 word piece blog article that I wrote about it, but it did really kind of strike me as one of the first instances of a really deep community wide discussion about a single prominent public piece of visualization work.
GEORGINA:
Brilliant, yeah, just because a piece of data visualization is popular and widely kind of reproduced doesn’t necessarily mean that it is the best way of representing that data and like you say, people might have different interpretations of that.
ANDY:
Absolutely and sometimes when asked to think about it from an like an inverted perspective of is it the least worst way to show something because things are complex you know some analyses are inherently complicated technical or complex and I think if we smooth those edges too much you lose the essence of a piece of analysis or a story and so sometimes it might take a while it might not be easy to read and make sense of but that might be okay it might.
be fine just to put a little bit of legwork in and work out how to read something. sometimes that inverted perspective is an important way to sort of, I guess, look at things in a slightly more mature way.
GEORGINA:
And if we could just backtrack a little bit, this might be a bit of an obvious question, but why would a person use data visualization? What’s the goal of it, as opposed to just putting in, say, a plain table of numbers?
ANDY:
Yeah, and I think it’s important to recognize we still need tables of numbers. We still need facts, figures, statistics. If the purpose of your communication or your presentation is to enable numbers to be read back or in a table, look up X and Y, like a bus timetable, for example. But if you are wanting to enable the viewer, let’s call them the viewer, the reader,
If you want to enable them to see it and to see it terms of its collective relationship, the patterns, the normal, the exceptional, the trends, the big and the small, looking at data through the values alone will not help you to see that. You can look at it and read it, but you can’t see it. And so what we’re trying to do is add value to the data to unlock new ways of understanding the relationships and features that inherently exist, but that are just really hard for us to…
when it’s just numeric or in text form alone.
GEORGINA:
Yeah. And data visualization is not a new invention, is it? I mean, there’s a lot of new sophisticated stuff going on at the moment, but there are some quite early examples of it, which I think are quite fascinating. Could you talk about those a little bit? Are there any that are your favorites from the kind of the very early days of data viz?
ANDY:
Yeah, I guess the authors of books when I was getting into database would focus on things like Charles Joseph Menard’s, the Napoleon’s March piece. There’s the cholera map by John Snow. There’s the Florence Nightingale, Cox Coon plot. These are some of the of classics that are always, you know, sort of recognized correctly. So as some of the most
pioneering early methods at times when production of these things was was quite manual quite primitive personally I I guess I have two of those you know kind of heroes from the sort of 1700s period which would be JB Priestley who developed the it was called the Priestley timeline he did lot of analysis of kind of historical biographies and
life spans and who was alive at the same time as other people and he was also born about four miles away from where I was born so I have to pick him as well in West Yorkshire but there’s a lady called Emma Willard who did these incredible timelines of things like the Roman Empire but she portrayed them almost in a in almost like a sort of biblical dramatic sort of
3D perspective form and sometimes she created these sort of almost like temples and the complexity of the production of what she made is just off the scale. And I guess I pitched those two also because I am a sucker for things that plot things over time. I think for me that is the most quintessential form of storytelling or story forming from our perspective as the reader and so I will always obsess over those kind of things that tell me how something’s changed over time or when things have happened over time.
GEORGINA:
It kind sounds like you’re saying that art itself is sometimes like a painting, can be a form of data visualization if it is trying to convey a lot of information to you at once.
ANDY:
Absolutely. this is where, in the early days of my involvement in the field, a lot of the most contentious discussions were about where’s the boundary between, let’s reduce it to simple terms, a chart and art. Where do those two things collide? Where do they overlap? When does one stop and when does one start? Is data art still part of data visualization? You’ve used data as the material to form
shapes and and patterns but your motive is not maybe to inform but to inspire to engage on an aesthetic level so yes I think it’s certainly part of visualization artistic elements and data visualization can be artistic as an output as well and I guess as with art one’s judgment of its relevance in a data this context will come down to one’s own opinion of whether they get something from it or not.
GEORGINA:
And in recent years, like we were saying, data has kind of evolved a bit quite rapidly. As we were talking before the recording, you mentioned the phrase golden age of data visualization. Could you tell us a bit about about that evolution in recent years?
ANDY:
Yeah, and I think where we are now 2026, just checking off credit, yeah, that’s it. It feels to me about 10 years ago actually, or around that time. And I call it a golden age because it just felt at the time that we had all the major, let’s say media, newspaper platforms doing brilliant stuff, huge teams, prolific output of work.
still innovating, still finding new ways of engaging different audiences with different topics. You also had a terrific influx of new talent in the field. Practitioners who may not be working in it full time, they’re hobbyists, they’re enthusiasts, they’re part-timers. You had also this kind of, I guess, the peak era of social media before it sort of demise as it’s kind of now.
But certainly in that area you have a very kind of vibrant and hopeful and positive social environment where these things would share like wildfire. But I also think one of the major milestones in this field around that time, probably maybe a year or so earlier, was the work of people like Georgia Lupi and Stephanie Posivec and their project called Dear Data, which for many, many people they can trace back their spark of, what’s this?
data visualization thing about. And Dear Data was a project that Stephanie and Georgia worked on to send each other hand-drawn data-driven postcards each week over a year based on a topic of their selection. So it might have been, many times have I said hello to somebody that week? And they do some sort of ornate, unique plot. Or how many times have I sworn? And maybe the severity of the swear words. But it was all these hand
drawn but personally collected data sets that in many ways democratized data viz more than any tool had done before that because you can now do popular data viz with pen and paper and you could do it on your own data collection not on something that was just a public data set from an agency or organization and I just feel that you can chat to people in the field now and many were inspired by that sense of a lower barrier to entry. And so I think that that period maybe through to, and possibly including the pandemic frankly, because it was all hands on deck data, data driven reporting. I feel that that was the sort of five year period-ish that you could pin down as a particularly golden period in contemporary database.
GEORGINA:
Yeah, I remember on Twitter, it was just constant really that people were putting such beautiful data visualizations up during that kind of time. when I really got into it was sort of 2018 and then up until the pandemic really. And yeah, there was so much stuff coming through. Now we’re obviously in to the age of AI and that has been helping people to kind of up the stakes a little bit with data in some senses. So you can use AI tools to create data visualization in the first place. Whether the data is correct is up for debate, but it can also make your visualization 3D. It can cover it with chocolate sprinkles or what have you and make it look really professional. Do you think AI is helping people to create data viz or is it maybe just tempting us with gimmicks?
ANDY:
this there’s a huge part of me that wants to be the more kind of hopeful not old man in response to this question but I think for now I might just let the old man speak more loudly because I do think a lot of it at this point in time or at least a lot of the practices we’re seeing at point in time are on the kind of gimmicky end of the spectrum and maybe even slightly more it’s not danger in a real sense but going beyond that as you said the data
being fed into these things is very contentious if it’s sourced just by AI without human intervention. I’m not sure if you look at lot of the AI generated visualizations if you can truly say that they are accurate and portraying data in the correct way. for example, if you were doing a bar chart in AI, sometimes I’ve seen these produced where the bars do not represent the numbers that they’re supposed to.
because you’ve not been hands-on in the production and you are therefore not able to take responsibility for every mark that’s left in that visualization, it does leave me with a degree of concern about the practice now and going forward. But putting the old man shouting at the clouds aside, I think there are many ways AI can assist, not take over, but assist us with tasks.
in the process that leads to a data visualization being produced that might be time consuming, that might be very manual, that might be something that’s not scalable or repeatable. I think that’s absolutely where we can get assistance. And also, on the elements of a data visualization that are not bound by necessarily objective do this and don’t do that rules.
For example, know, coloring a bar chart bars, unless you’ve got a corporate identity that you’re trying to follow, unless you’ve got certain, let’s say, visual accessibility concerns that you’re wanting to sort of preserve, the specific choice of those bar coloring-wise is up for grabs. It’s subjective. So why not allow some kind of AI agent to suggest some different tonal qualities, some different kind of combinations that work nice together?
I think where you’ve got these subjective moments in data, and there are lots of those, we can have AI inspiration. Where you’ve got lots of tasks that are grunt work jobs, let’s have that AI assistance, but still with the possibility of us just checking things and intervening and refining where it needs to be. So I am not completely dismissing it at all. I think there are loads of ways it will help us incredibly. I just think as a craft,
there are too many parts of it that have rights and wrongs that an AI currently, from what I’ve seen as a practice now, isn’t properly demonstrating in the things that it’s producing.
GEORGINA:
Yeah, I think I’m also on team Old Man with the actual building of a chart side of things, but I agree with you that AI can be a useful partner, let’s say, in helping with the design process. On design, this is something that you have written about extensively. It’s kind of the theme of your big handbook on data, Viz, is really about design. Could you run me through the fundamentals of data design as you see it.
ANDY:
Yeah, I mean, and the word design itself, I think, can sometimes…
It can sometimes give people an impression that this is a subject that’s not for them. I’m not a trained designer. Many people in the field that I know of are not trained designers. But we are designing as an act. And it is something that you can learn through practice and exposure and inspiration. But I think if we look at this topic through two perspectives, first of all, the parts of design. What are the components of a visualization that have this element of design? There are five.
There are five components of the design anatomy. You’ve got the data representation element, which is essentially the charting, how you depict the data through different marks and attributes. You’ve got the potential interactive design. So if you want to, or you’ve got the option to make it digital and interactive and interrogatable and modifiable, maybe animated, that’s an option.
We’ve got to think about what we call the annotation layer, which is elements of the sessions, like titles, labels, scales, legends, all these relatively on the face of it mundane pieces of apparatus, but vital to help the reader know what they’re looking at and how to read it. You’ve got the color choice. Now, the colors are both color associations, like categories. Think of the London Tube map. We associate different line colors with different routes. But then there’s also a decorative quality.
of coloring the grid lines in the background that we’ve to think about. And then lastly, what we call in a design context, the composition is the arrangement of all this within the space, the screen, the page that we’re working to. How do we size it all, fit it all together, arrange it on the page? So these are the distinct design elements that we’re facing. But alongside that, we’ve got to think, well, what is good design? And in my opinion, although it’s very much influenced by many other authors in the past, good visualization design is trustworthy. So all those choices around the anatomy I’ve just mentioned there need to make the visualization as reliable, as accurate, as fatal to the subject and the data as possible. It should be accessible. And this is not just about visual accessibility, but it’s about broader access to understanding. So can you help me to understand what it is I’m seeing? Can you help me to understand what’s important, less important?
Can you piece this together in a sequence that makes sense as a sequence of understandings towards an outcome? And then thirdly, it should be elegant. Now, it is the least important of the important principles, but why make something horrible is you can make it beautiful. So this is the most subjective of all those principles, but we can still aspire to make our work refined, appealing, tasteful, and just fundamentally joyful.
Those are the elements of design that I try and tackle in the book, the principles but also the elements that make up the anatomy.
GEORGINA:
And those elements of a chart design can really influence the message that’s communicated from it. I know that from having looked at quite a few data visualizations. I’m thinking of one in particular, I think you’ll have seen as well, which is a chart that appeared in, I think, newspaper of deaths in the Iraq War. This was a few years ago now. And this was a bar chart.
and the bars were pointing downwards, whereas normally a bar chart, you’d have them pointing up. The color of the bars was in this kind of blood red. And so it had the effect of kind of like, as if blood was dripping down the page as these bars were kind of mounting up. And so that chart was to convey a sense of this is terrible, that there are all these deaths occurring. Somebody then did an exercise where they…
flipped that bar chart the other way up, so conventional way up, so you could see actually the bars were kind of rising, but then they fell again and they changed the color of it from this blood red to a kind of mellow blue color and the message was completely different. So they illustrated that using this exercise basically. The message was then deaths are on the decline and it was exactly the same data, but just a flip chart and different choice of color.
So from your experience, I’m sure you’ve got lots of examples like this. Is there anything that you could say is kind of like the most important thing about the aesthetic design aspects of a chart?
ANDY:
I mean, think, I mean, this is perhaps the very hub of the whole subject in many ways, but I think the most important thing there is…
no visualization can ever be truly objective it’s always the consequence of the choices that we make. it going be red? Is it going to be light blue? Is it going be going down? Is it going be pointing up? There isn’t a single right solution only your solution and I think sometimes when I speak to a lot of people who are sort of new into database they sort of that sort of responsibility that comes with having to make these choices but I think
We have to embrace the role of being editors of our work and take the responsibility to say, I am responsible for every mark, every decision that’s linked to this piece of work. And so I’m going to think about every decision carefully. I’m not going to do things randomly or arbitrarily. I’m going to think about everything and know that this will be a certain consequence of those choices. So going back to the upturned inverted red bar chart, it clearly has the metaphor of dribbling blood, as you mentioned.
it perhaps loses the sense that deaths are improving because they’re shrinking but it does convey that lots of deaths have happened when those happened is almost irrelevant because they’ve happened whereas the the alternative version by Andy Cotgrieve is showing a more like a timeline in sense of things are getting better so each of those has different choices that are both legitimate they’re both
the they are both legitimate perspectives of the same truth but looking at it from a different perspective and that idea of perspective is something that we have to embrace as I’ve said we are the photographers of this data and we choose when to take the photograph what angle to take it from what to leave in what to leave out but when we make choices like making a bar chart red the connotations are clear the contations of color in certain cultures are clear. Color is perhaps the most powerful visual cue out of all these things that we choose, whether it’s the chart or not. It’s the first thing that our brains sort of process. there’s lots of responsibilities on our shoulders. And again, I think once you embrace the idea that there isn’t a single solution, there isn’t a single right solution, there are just different solutions, it can be a bit more liberating to say, right, well, this is my solution that I will choose to take.
GEORGINA:
I love the idea of embracing the role of editors in creating data viz. Can you say in general, though, I mean, it’s obviously very hard for you to make big generalizations, but you see a lot of charts which are quite minimalist, quite simple and others which are quite overloaded with stuff. Can you say in general whether minimalism and simplicity is always kind of preferable?
Because it sounds like we’re talking about data visualization that tries to convey a message. It’s trying to communicate something. Is it better usually to keep that simple, or does it depend?
ANDY:
I mean the most boring two words in data vid are, it depends, so I’m to avoid those two words and I’m going to say for me what’s more important is clarity rather than and we can get into semantics of language here but the way that I see simplicity is it’s somewhat reductive you know you are removing and reducing things to make it either quicker simpler or easy and there are many situations when you want that when time is of the essence
and immediacy when the role that you’re asking of your audience to make sense of, you know, maybe multiple different channels of communication isn’t viable because they just don’t have that prior knowledge of how to make sense of these things. So everything is fit for purpose depending on the context that you’re working it towards. And I think once again, if you avoid
pursuing just the goal of simplicity and focus more on clarity a big each other can still be making that thing as clear as it can be indeed sometimes the big enough to watch out the chaos of our job maybe the story itself if we take for example politics which you know we all know it in many in many many countries politics is not a stable situation from one election cycle to the next this all sorts of flux and changes
and if you look at recent elections and the courage of those elections and you see charts where you’ve seen and all these people used to vote for this party then they went to that party and that party and you look at a chart you might think there’s chaos this is a bad chart it’s not a bad chart it’s a good chart because it’s showing the chaos it’s showing the fact that these things are not stable because if it’s stable it’s boring nothing’s happening which in itself can be a story so i think to focus on
pursuing the clarity of what it is you’re trying to convey gives you then the pathway to go down a simple visual route, quick, simple and easy, but also allow you to go down a more complex, maybe time consuming, maybe busy display because that still might serve the purpose of what you’re trying to explore.
GEORGINA:
Some people listening might be thinking, this all sounds great, but I could never do it myself. I don’t have the skills or the time or the vision, something like that. Now, I personally think that everybody probably has it in them to come up with a data visualization, but I get that starting out with it is daunting, both because of the data wrangling side of things and also that design process that we’ve just spoken about.
Now, in your book, Data Visualization, the handbook which I mentioned, you don’t dive straight into the technicalities of making data vis. In fact, you pretty much solely focus on the steps before that, so the decision making process behind it all. And you talk about the hidden thinking behind a data vis design.
ANDY:
That’s right.
GEORGINA:
Could you talk a little bit about what you mean by that and how that might help people to get started with DataViz?
ANDY:
Sure, earlier I listed the different elements of a design anatomy, the charts, the colours, everything else. They are the things that we see that manifest in the final piece of work. So that’s the visual thinking, the hidden thinking is all the preparation work that goes into a day to day design that isn’t the design. It’s the thinking about the audience and who they are, what do they want, what do they need.
What’s their situation? Are they under pressure, in a rush? Are they got a nice coffee, relaxed setting? It might be a more prolonged engagement. Have they seen this chart method before? Do they have an interest in the subject? Where’s the data from? Is it good quality? Is it trustworthy? What’s the shape, the condition? What’s the range of values? Because that might affect, let’s for example say, if you’ve got eight categories, that might lend itself to a bar chart. If you’ve got 100 categories,
that bar chart’s going get pretty busy. So you might need to try and find an alternative. The shape of the data has a huge impact, just as it is if you go to IKEA, you come back with all the materials, all the bits, you empty them out on the floor. Well, that’s the data. You’ve got different shapes and amounts and physicalities that will or will not fit into certain charts. You’ve got to think then about the different stories that are possible, different angles of analyses. Are you going to show how this has changed over time? How it looks?
geographically how it breaks down by category or all the above? Have you got space for all the above? If you’re going to show all three, what’s first? What’s most important? So none of these things are design-led thoughts. They are contextual, they’re analytical, they’re editorial, they’re inherently human. But if you don’t do that clarity of thinking, when you get to the design stages of your process and you’re working out what colours to use, what layout to follow,
you are having a basis on which to make those decisions effectively. So the hidden thinking is all the hidden part of the iceberg that we’ve got to go through first before we start to make these design choices. Now, anyone could do that. Even if you’re not an advanced Excel user, R user, statistician, if you can count things, if you can categorize things, if you can group things, if you can do things just as simple as averages.
ANDY:
maximums, minimums, then you are able to collect, capture, make sense of data and then plot it. And I spoke earlier about the DIA data project, this artisanal hand drawn data collection exercise and data designer exercise. Pen and paper, if you can pick up a pen, you’ve got paper, if you’ve got colours, you can create these things yourself. I think the most important attribute that one needs to demonstrate is being curious.
ask questions about things, wonder about things. How many times have I taken a drink of water today? Am I drinking enough? I’ll start to plot it. And when do I drink water? When do I have the longest period when I’m not drinking? These are all things that are driven by one’s own questions about things. And so getting into the habits of questions, questions, questions, wondering what’s the answer to something? What’s the pattern behind something? That’s the most important first step.
GEORGINA:
So you talked about somebody with a pen and paper coming up with a data visualization, but in practice if they wanted to do this for the company they work for, for example, they’d need to be doing something probably more sophisticated than that. You mentioned Excel. A lot of people are probably somewhat familiar with it, but not experts.
Now let’s say somebody has no real experience even with Excel or with statistics and certainly not with other statistical software. Is it possible for them to still create data viz on the computer basically or on their phone perhaps or is it unrealistic to say that this can’t be done without improving some technical skills?
ANDY:
I think ultimately the answer would be you’ve got to have at least a surface level appreciation of a spreadsheet and specifically a table of data. If you can organize your data into a table of data, many of the tools that are out there, now there are hundreds of tools out there, which is kind of crazy in terms of a field having so many different options. But if we look at some of them all, perhaps
common popular and accessible tool options and we can add these into the show notes later on but Flourish is a tool Raw, R-A-W, Raw Graphs is another tool and Data Wrapper there’s three tools that would always recommend to people because they’ve got tremendous free versions now Flourish, Data you can pay for as well
But they’ve all got free versions that allows you to use them to create charts using the browser. And they’ve got different levels of sophistication. But if you can start off with a table of data, you can copy that data and paste it into each of those three, or least connect it to the spreadsheet in each of those three, and create many different charts, many different designs of those charts. In the case of Flourish and Datawrapper, many different interactive versions of those charts. Without the need to code,
with at least plenty of access to tutorials, but even then the user friendliness is quite clear from each tool. I would always start with those three because if you can just get to grips with what those three offer, you’re able to make loads of different charts. You’ll have a wide range of ways of expressing data visually, but you’ll also just have this excitement of, wow, I could make that thing. And it even might be a chart that’s not
one of the common three, the bar chart, the line chart, the pie chart, you might start to have things with bend your lines, stream graphs, bump charts, alluvial diagrams. So that excitement just then gives, as with anything, when you’re learning something, once you get to a staging post, think, wow, I’ve done that. It just gives you that encouragement to move on to the next level. So just a surface level appreciation of how to make a table of data in Excel or other spreadsheets, you’re in. You’ve got a chance of using the technical tools that are out there.
GEORGINA:
Yeah, really great recommendations for those online tools and I will put a link to this but on your website you’ve got a really nice kind of table of well like a grid of different resources that are available online. I think probably most of them are free and yeah we’ll just take a little bit of learning to get to grips with but they can help you to produce practically any type of chart that you want to so I’ll definitely put a link to that.
I did wonder though if somebody makes a chart and it just looks rubbish, do you have any hacks that can make a data visualisation look instantly a bit more professional?
ANDY:
I that perhaps the best way, and now we talked about simplicity and perhaps focus on the clarity, maybe this is when simplicity does come back into the thinking because I think when things look a bit garbage it’s usually because they’re full of grid lines and tick marks and labels and they’re weighed down with all this apparatus that isn’t necessary. So my first thing will always be strip a chart back.
get rid of all the labels, rid of all the grid lines, all the tick marks, and just leave it as the marks and the data values that they represent, whether it’s coloured or different heights or positions or lines. And then gradually add things back that you think are relevant. If you’ve got 10 bars, maybe there’s only one of those that’s actually that important to label to say, here’s the key story, this one here. Think about the title. Don’t accept any defaults. The title probably
in some cases can be here’s what’s plotted x vs y over time broken down by what ever but maybe think more ambitiously about a headline what’s it saying what’s the key takeaway you want to get from this or maybe it’s a question that the chart answers so think more creatively about your your chart titles the colours once again strip it back to effectively sort of black and white and greyscale and then add colours only
when you need to associate something with a certain colour or whether you want to establish a colour ramp for a sort of lighted out shade for colour values or whether it’s something where you’re trying to just highlight certain values and so the contrast if you’ve got 10 bars, 8 might be grey, 1 might be of one colour, 1 might be of another colour because they’re the two that you want to focus on and draw attention to. So this is perhaps, know, this sort of journey of simplifying at this point I think is most relevant. Strip it back and start to challenge the need for every one of these component elements.
GEORGINA:
Nice. So start with bit of decluttering and then add things in as needed. So we’re going to have to wrap things up fairly soon, unfortunately. But you look at a lot of data visualizations all the time, obviously. Is there anything that you’ve seen recently that you thought was really great, really innovative, or anything that you’d like to plug?
ANDY:
I will always like to plug in some things, I’ll come on to that in a second. I think at the moment, so I do try and keep on top of the latest, best things in database and I think I’ll pinpoint a couple, I think first of all in terms of just a constant stream of inspiration and brilliance, writers’ graphics are exceptional.
everything that they do is wonderful but it’s also just different it’s always the right solution in my mind for what the story is whether it’s the colour scheme whether it’s the methods of mapping so go onto the writers I think they’ve got a visual gallery I can’t remember the exact URL but there’s a gallery of their visual work and it’s just terrific stuff in terms of new things that have emerged I really like a new
newsletter from Amanda Shendruk who used to work for the Washington Post and before that for Quartz I think it was Amanda’s got a great new newsletter called it’s called Not Ship and basically every week she focuses on a sort of data-driven story or a data-driven investigation it’s almost like she’s still practicing the art of journalism, visual journalism but she’s now independently doing that so Amanda Shendruk
has a eye for a story, brilliant eye for different visuals. They are two of the best things that just fill me with visual nourishment every month. I think the one thing that I would plug in the next month or two, you mentioned before the Seinfeld project that I did a few years ago now. Well, maybe next month, hopefully next month.
there’ll be a new version of this coming out which is a digital version so the original version was just a book and I did a limited print run to match the number of episodes 176 and I sold the book on a pay what you can basis and all the proceeds went to charity but only 176 people effectively saw that project so I thought I need to widen the audience a little bit so there’s a new version coming out next month where we’ve made it like a digital scroller telling piece but also there is a lot of sonification involved
ANDY:
So we’ve taken things like, imagine an episode, and I captured data about when the characters were on screen together, and for how long, and in which scene, and in which location, and who got the laughter. We’ve turned all that data back into a sonification where we assign instruments to different characters and locations, and then create a brand new, unique piece of jazz out of the data. In some cases, it sounds horrendous but it’s a fun, not important piece of work. So that’s coming out next month and hopefully more than 176 people will see it.
GEORGINA:
I know what’s going to be on my Spotify wrapped at the end of this year then. As my final question, I just wanted to ask, and maybe you don’t have an answer to this, is there something that you’d like to see visualised that you haven’t seen yet? Or I suppose put differently, is there anything that you’ve really tried to visualise, to do a data visualisation on that you just haven’t been able to make a reality?
ANDY:
It’s a really good question. don’t think… I think there are certain things that I’ve worked on that are harder than others. think showing things like… Showing things like how networks evolve over time is a really difficult challenge because you’ve got a time dimension, you’ve got all these complex air balls. And I’ve had a couple of projects that never can I go off the ground because…
all we were ever really able to show was complex hairballs and it didn’t really add any value. So that’s a scratch that remains yet to itch. I think what I’d probably do is answer the question more broadly, but bring it back a little bit into the AI discussion because I think one of the things I’m most hopeful about with respect to AI is its role in helping us to unlock more qualitative data rather than quant. And I think things like feelings, opinions, just even categorizations or classifications of how to reach, how to access, how to mine data. Even if it’s not perfect, will be more perfect than what we can do as humans for a lot of these things. So to my mind, that is the next great opportunity that we’ve got to tap into these softer edges, these more elusive data structures that
and you know going back to the side of a project i’d try to classify things by watching the episodes if i could have introduced a an assistant that helped me to do all that it would have not just give me a quicker route to get the data probably on lots more things like the joke topics or the the kind of the relationship categorization so these are things i am really excited about the prospects of going forward to just to give us a whole new sort of well of data prospect.
GEORGINA:
Super interesting. Thank you so much, Andy, for talking to me today and coming on the show. My guest this week was Andy Kirk, expert in data visualization. If you’d like to learn more about designing powerful visualizations and develop your own technical skills, take a look at the training courses Andy offers on his website, visualizingdata.com. I’ll be back next week with a new guest.
ANDY:
My pleasure, thank you.
Originally published on Substack: https://georginasturge.substack.com/p/the-power-of-data-visualisation-with