Data visualisation is both an art and science!
What I hope to do with this post is briefly explain why I think this and why visualisation is important and provide you with some additional resources that comprehensively explain the topic as well as provide practical day-to-day support for your data visualisation endeavours.
We will navigate this topic by discussing each of the following sections:
- Firstly, why both an art and science?
- But why is it important anyway?
- So what makes a good data visualisation?
- Practical data visualisation tips (not an exhaustive list).
- Looking for a quick start guide?
- Casual blog content could be your thing!
- Looking to Deep Dive?
Firstly, why both an art and science?
Well my view is that there is a body of knowledge surrounding how the human eye and how humans perception of the world. For example, humans exhibit the following characteristics:
- We are more likely to notice a moving object rather than one that is stationary.
- We are not great at comparing angles
- Our eyes have limited bandwidth
- We are great at predicting straight line trajectories but not so great at exponential
As such the field of data visualisation is founded in this science, both intentionally and unintentionally.
It may be surprising to some that data visualisation efforts have scientific grounding – it has been reasonably well documented by Edward Tufte and others – more recently Stephen Few from the Perceptual Edge covers this fairly comprehensively in his books “Show Me the Numbers”, “Now You See It” and “Information Dashboard Design”.
I think it is also an art. I would say that most practitioners don’t have very clear understanding of all the visual science nor have a well structured framework with which to apply in an organisational setting. In the day to day, I think practitioners consciously apply a limited set of rules they have learnt that are grounded in scientific principles. However they also rely on intuitive feeling and in some ways this is probably driven by sub-conscious pattern recognition (i.e. having seen what looks good and works elsewhere and by building a pattern model in their mind, they apply this in new situations)
I think most people can appreciate why it may be perceived as an “art” given it’s nature as a visual medium.
So yeah, I think it is both a science and an art.
But why is it important anyway?
As a means of communication is extremely valuable. Humans have 2 eyes and 1 nose and 1 mouth/tongue and key sensory organs their quantities are indicative of how important those senses are. What I mean is that we have 2 eyes because our vision is so important to us. Having 2 sensory devices for vision also plays a part in the speed and bandwidth with which we can absorb information from the environment around us.
So I am hoping you are able to appreciate that data visualisation is therefore an extremely important means by which we can receive and/or communicate information. I think this has been succinctly summarised with that say “A picture is worth a thousand words”.
We should also recognise that information transmission or communication via data visualisation does not equate in any way to computation or our ability to think and author complex ideas / process information we have received. Data visualisation is a communication medium for which we can transmit those thoughts.
So what makes a good data visualisation?
There are a lot of opinions on this topic, however when trying to distill it down a good visualisation can make such a difference on whether the published content is used, continues to be used and the value that it provides the consumer. In some ways this is similar to any publishing activity and particularly there are parallels to social media content authoring.
The factors that affect it’s consumption include:
- whether a person is “enticed” or initially engages with the content.
- whether the content tells a story and hence continues to engage.
- whether the read continues to come back to the content
- with modern reports and especially dashboards, it is important to be able to communicate meaning without becoming noisy and cluttered.
- there is also the challenge of being able to communicate a density of information without losing consumer engagement.
Practical data visualisations tips (not an exhaustive list):
- Keep it simple, but not simplistic.
- Treat the process like sculpting. Remove everything that does not communicate meaningful information. So this means where it makes sense, remove borders, unnecessary pictures, unnecessary colour.
- Choose the right chart for the right story/situation. For example use line charts when trying to demonstrate a trend over time. Several reference guides are listed further down in the blog on this topic.
- Avoid using pie charts, especially when there are a lot of elements as it is a bad use of a rectangular surface (monitor or paper). Humans are also not great comparing angular objects and so we end up having to put data labels and this inevitably makes the report/dashboard look busy.
- Be careful what you emphasise. You really should be focusing the consumer’s attention to those things that matter. For example, be careful about over-emphasising a chart title. The focus really should be on the data in the chat. Chart titles should really be a compelling and brief narrative on the chart content.
- Following this thought, be careful where you use text as this is not an efficient means of communicating information and should be used in a very targeted way.
- Use colour primarily to communicate something meaningful. An extreme example might be to ensure all elements in your visuals are by default grey and use red to signify exceptions and alert users to breach of some threshold.
- When using colour, it is best to maintain consistency in it’s application. For example you might choose blue to represent budgets or targets. You should keep this use of blue consistent throughout your reporting and visualisations.
- Shapes – in the same way that colour can communicate meaning, shapes can also efficiently do the same. So use shapes, again consistently, to provide contextual understanding for the story you are trying to communicate.
- Use predictable layout patterns. We feel calmer and find it easier to navigate a report or visualisation when we can predict the layout. For example this might mean largely maintaining symmetry with visual object layout.
Here are some additional links to data visualisation tips (some of which are already included above):
- Data visualization tips for clear communication | Geckoboard
- 10 basic data visualization tips designers should follow (bigdata-madesimple.com)
- 25 Tips to Instantly Improve Your Data Visualization Design (columnfivemedia.com)
- 6 great data visualization tips for more effective and engaging design (tableau.com)
- 5 Amazing Tips for Data Visualization | by Shelby Temple | Towards Data Science
Looking for a quick start guide?
Below are some visualisation guides:
The Data Visualisation Catalogue. This is an interactive website for choosing a visualisation.
Data Story Visualisation: A Decision Tree (link) by Stan Pugsley. Note that this is not just a “visualisation guide” but a complete blog explaining the guide.
The Visuals Reference by SQLBI (Power BI focus)
There is a particularly good guide if you are working with Microsoft’s Power BI data integration, modelling & visualisation software.
Casual blog content could be your thing!
In addition to this perhaps take time to look at the following blogs (this is only a sample of what is out there):
- What data visualisation experts wish they knew when they first started by Evelina Judeikytė
- God and moses? The difference between Edward Tufte and Stephen Few by Jorge Camoes
- The Perceptual Edge blog by Stephen Few
- The Functional Art blog by Albert Cairo
- Visualising data
- Storytelling with data
- Driven by data
- The Data Visualisation Catalogue
- Information is beautiful
- Spurious Correlations (also a statistical blog, what not to do)
- Flowing data
- Modern data
- Tableau blog (well known for it’s visualisation capabilities)
- WTF Visuazations
- The Economist – Graphic Detail
- Junk Charts (what not to do)
- Deconstructing metrics by Conor Dewey (Towards Data Science)
- The Data Visualisation Corpus (on Kaggle, kaushiksuresh147)
- A Beginner’s Guide to Data Visualisation: Aventis Learning Group
Looking to Deep Dive?
If you have had the interest and/or time, you most likely have come across the names of visualisation experts such as Stephen Few, Albert Cairo and Edward Tufte (just to name a few).
Come to think of it, there is no shortage of books on the matter. A quick search of Amazon returns a list of over 2000 books (Amazon.com : data visualisation books).
This might feel a tad overwhelming. Where to start?
I would certainly recommend reading Stephen Few and Albert Cairo books – I am certain there are plenty of other good reads – perhaps you can leave your suggestions in the comments or on the Lucid Insights LinkedIn post relating to this blog.
Stephen Few books
Albert Cairo books
Do I get it right all the time? No, but improvement comes with practice and feedback.
Do I think I am an expert? No but I think I am pretty close. I continue to learn though and I find it interesting.
Do others think I am an expert? I would say yes. This judgement is a relative concept.
Is data visualisation the be all and end all of data analytics? Definitely not, there is so much more in this field, but it is an important area to keep in focus.
Where did you pick up your data visualisation skills? From experience as well as reading some of the above content mentioned in this blog. I also keep in mind, whenever I see someone else’s data visual, to note the good and bad elements. I also look at other “fields” and observe any elements that may translate across into the analytics / data visualisation space.
Are you sure we can’t use pie charts? Of course you can use pie charts, it is just not the preference by default. It just depends on the circumstance and your audience. At the end of the day I think people in general find comfort in the organic shape of a circle. Just be careful where and why you use them.
I hope you enjoyed this post!