Data Visualisation 2 – because there is so much to Data-Driven Design
This month I read Andy Kirk’s absorbing Data Visualisation 2, or to give it its proper title “Data Visualisation” 2nd Edition.
The subtitle for this book is ‘A Handbook for Data-Driven Design‘, that hints at how this is packed with advice. Although the paperwork version is a comfortable weight it is astonishing how much it contains. This really is a guide for practitioners and one they will want to refer back to.
Comparing this book with two other Data Visualisation books that I’ve previously reviewed might be a good place to start. “How Charts Lie” by Alberto Cairo is much more aimed at a general audience, all concerned citizens if you like. “Storytelling with Data” by Cole Nussbaumer-Knaflic includes more on the wider challenge of storytelling. Plus, it narrows its focus to fewer chart types and a design style that works in most corporations.
So, I was glad to find that Andy’s book took me deeper & wider into the world of data visualisers and all they need to consider. If you have a responsibility to visualise data in reports or the results of your analysis, I recommend you take the time to study this book.
Laying a firm foundation
Andy start’s this book with a thorough introduction to the contents and directions to the online resources that accompany it. These provide an opportunity to see more data visualisation examples, access related resources and test your knowledge with practical exercises.
This hints at the style of this book. It is perhaps the best example I have seen of what should be a textbook for those learning this craft. So, if you like to read a book cover-to-cover (like me), before later dipping in for reference, don’t expect an easy read. But you will learn so much.
With regards to your ability to later dip back into this book – it is well designed to achieve that. Each chapter ends with a consistently presented, short but complete summary of the key lessons learnt. Sufficient to prompt memories or guide your revisiting.
Andy is also the kind of man to walk his own talk. Consistent use of colours for chapters and his workflow model also aid navigation. There is also a pleasing symmetry to the consistent use of all design elements at the start and end of each chapter. Plus I’ll come to the “treat” in the centre later.
Walking you through how to do it
The structure of this books is chronological, or in other words, walking you through the design process of a visualiser. There is too much important content for me to do justice to it in this review, but here is a very high-level summary.
Defining Data Visualisation
In the first chapter, Andy unpacks one of the most useful (whilst avoiding being verbose) definitions that I’ve heard. He defines Data Visualisation as:
“The visual representation and presentation of data to facilitate understanding”.
We learn why we need to be think about each element of that sentence.
The Data Visualisation Process
Here Andy introduces a workflow of four stages:
- Formulating your brief
- Working with data
- Establishing your editorial thinking
- Developing the design solution
As well as explaining why each stage matters and what will need to be considered, he also introduces a number of guiding design principles. This is particularly helpful for those analysts lacking an art/design background.
He recommends 3 design principles of good data visualisation, that good data visualisation design is:
Formulating your Brief
In chapter 3, Andy outlines what to consider when formulating your brief. This includes plenty of practitioner advice on topics including context, stakeholders, constraints and deliverables.
From this chapter onwards you also enjoy more of the breath of Andy’s examples of good data visualisations. One of the reasons this will be so enjoyable to dip back into is the practical examples included.
Working with Data
For my more Analytics-centric audience, chapter 4 will be right up their street. Andy steps the readers through most of the pitfalls when acquiring, examining, transforming & exploring the data you will use.
As I find on my own Data Visualisation training course, it is so important for analysts to realise there is data prep needed for Data Viz too. In fact, it may prompt other data quality & provenance checks that should have occurred prior to analytics work.
Lots of useful tips in here, including the important role Data Viz, can also play in Exploratory Data Analysis, not just the explanation/exhibition to stakeholders.
Establishing your Editorial Thinking
This is a step too often overlooked and Andy explains clearly why it’s needed. This chapter brings to life all the subjective thinking that is needed to create relevant & effective data visualisations for specific topics & audiences.
Taking time to consider your overriding curiosity and what is needed in terms of angle, framing & focus brings much greater clarity. The practical examples he includes bring to life how easy it would be to ‘go off on the wrong track‘ without taking sufficient care at this stage.
Developing your Design Solution
This is what feels like the fun part. After a useful (and impressively brief) summary of visual encoding, Andy provides his own categorisation of different chart types before he includes a beautiful central section.
In this book, he uses the CHRTS acronym to catgorise:
- Categorical charts
- Hierarchical charts
- Relational charts
- Temporal charts
- Spatial charts (mostly maps)
Then comes the central ‘treat‘ I referred to earlier. On a calming blue (rather than white) background, Andy includes 49 detailed examples. Each page introduces one important chart type. These pages are worth the cost of this book alone as they are such useful reference pages.
For each chart he includes:
- Alternative names for that chart type
- Description of key elements of that chart type
- A positive example (middle third of page)
- Presentation tips (annotation & composition)
- Variations & alternatives
It is tough to overstate how absorbing these pages are to those interested in which chart type to use. Andy closes this chapter with including factors and consideration to guide your choices. These include a link to the brilliant Chartmaker Directory that Andy publishes to help you see which software supports you in producing which chart types.
In this chapter, he summarises a number of features of interactivity you might use. Each is supported by considerations for effective use. This is more important to a wider audience today, as more of the mainstream BI packages support elements of interactive output.
The features considered here include: Filtering; Highlighting; Partcipating; Annotating; Animating; Navigating. Lots to consider & useful examples.
Another topic that is tough to explain (or master) without practical experience, but Andy expands our design thinking beyond what you might normally consider. In addition to the use of chart labelling & captions (which is what I imagine when I see this word), he reviews the use of:
- Headings & instructions
- User guides (especially if you have interactivity)
- Reader guides and legends
- Chart apparatus & references
- Footnotes & methods
Once again, as Andy does in all chapters, he also includes ‘Influencing Factors and Considerations’. This part of the summary is followed by ‘General Tips and Tactics‘. Considering all of these will add much more care to the words, fonts & other elements you may be overlooking.
What a delight to have an English author so that word is spelled correctly, ha ha! Ok, now that I’ve alienated my American English speaking audience, let praise how well Andy sets up this subject.
In similarly impressive brevity to that achieved for Visual Encoding, he summarises Colour Theory. His use of the HSL categorisation provides a useful way to think about your colour choices.
Through a wider variety of chart types than you might expect, you are then walked through the importance of effective choices for:
- Data Legibility
- Functional Decoration
These are explained with some beautiful examples to bring the potential impact to life. Together with those influencing factors/considerations and genera tips and tactics.
Chapter 10 closes the walkthrough Andy’s original 4 stage workflow. Akin to focusing on Editorial Thinking, it is a stage too oft-neglected by analysts. Most of those I have seen working on presenting data in today’s businesses would benefit from the guidance Andy provides here.
Before considering your design work complete, Andy explains the elements of composition to consider. These are all in order to achieve seamless visual journey, an experience for the viewer that will instil confidence (and perhaps trust).
This chapter considers:
- Quantitative value ranges (inc. outliers & log scales)
- Editorial thinking (yes, again)
- Data representation
- Elegant Design
Epilogue = the actual development cycle
It is perhaps a useful positioning of whom this book will help to say that it is only the epilogue that engages with developing your design in software. I am a fan of that. I see too much emphasis for analysts on technology & a rush to coding/software, with insufficient thinking time beforehand.
Nevertheless, there are some useful tips in this section too. This development cycle is recommended & briefly explained:
- develop a concept design
- create a mock-up/prototype
- conduct testing
- refine and complete
- launch and evaluate
Data Visualisation 2 – should you buy?
So, perhaps the raison d’être of any book review is to help you decide should you buy this book. For many my answer would be yes, but it perhaps more useful for me to highlight the audience I believe it would help.
Those whose job is to create data visualisations or communicate data using charts/graphs. If you fall into this group, unless your experience is such that you should be teaching others, you should buy this book. Not only could reading it & completing the exercises act as a training course, but it will also be an invaluable reference for you.
Those managing analysts or roles who produce data visualisations. I’d also recommend this book to them. It is sufficiently focussed on judging what is effective for the end-user, to be a useful guide for you too. In fact, it would enable you to better challenge/review/test & develop your analysts in this regard. A key skill for data leaders these days.
Who shouldn’t buy this book? Well, I wouldn’t recommend it to those just interested in software & coding (the latest Data Viz package for Python etc). Nor is it a suitable text for the general public (for whom I would recommend “How Charts Lie” as mentioned earlier.
But most of those I work with would benefit by reading & having to hand this important handbook. Well done, Andy Kirk, this book and your wealth of online resources continue to be an important contribution to advancing Data Visualisation & better design.