Does anyone know any guidelines in creating interactive visualisations for the exploration of datasets?

I can find lots of material on why particular instances are good or bad, but the actual design process of designing an interactive visualisation seems to be a black art.

4 Answers 4


Your focus is not to create "an information visualization," but instead to help your users complete certain tasks or achieve certain goals. The design of an information visualization depends on what you want your users to do with the visualization.

(This may be why the cases you've read about are about whether the tools are "good or bad." The visualization exists to help users, not to exist on its own, and so evaluating one and designing one cannot really be separated from what it's meant to do.)

Design isn't a black art: it's a greyish art that you can learn about and practice. There are many ways of knowing your users, knowing what they want to do (and how they already do or don't do it), and to test how best to help them. Whether you use task analysis, interviews, case studies, paper prototypes, or something else, your goal is to help them. These methods are the same you'd use to design a non-visualization interface, because they're about people and tasks.

For example, in LiveRAC, graphs of web traffic were not really made to "be a visualization of CPU load data," as much as they were made to help site administrators answer questions about server traffic and identify any problems or opportunities with that. The paper on LiveRAC goes over how the authors identified who their users were, what the users needed to get done, what data could be relevant, and how the LiveRAC infovis was made to help.

If an infovis is a part of how you hope to help your users, take a look at the notes of some courses on visualization design for design considerations and best practices regarding things like colour, angle, readability, chartjunk, etc. Again, not a black art, but something you can read about and play with.


I used the following textbook as an undergrad student, and found it informative: Information Visualization Design for Interaction by Robert Spence

Here are a few key high-level concepts:

  • The way data should be represented and presented depends on the type of data and how it should be understood by the viewer.

For example, the way you would present population density and size respective to location might would very different from how you would present a public transit railway map. Both present information related to locations, but each has their own desired focus.

For the first example, you might use a true-to-scale map with colours to represent density and dot size to represent city size. With a railway map, true-to-scale accuracy might be abandoned for the sake of understandability and colour might be used to represent the different 'lines' of travel.

In both cases, they would be considered 'maps' but on inspection they would be very different things.

  • The closer you can get the viewer's perception (think "at-a-glance") to the desired interpretation, the lower the likelihood of mis-interpretation of the data.

For example, with a million pieces of data plotted in a scatterplot graph, the use of colour could be employed to display 'clouds' of same-attribute data points. If there are under 8 colours used in such a manner, this could effectively display patterns data. If there are many colours used, however, this would be less effective due to human limitations on how we perceive colour differences. Dark-blue data and purple data effectively become the same thing in a larger (10+) group of colours, mis-leading the viewer into seeing clouds of data where there is none.

The book goes into more detail on more facets of Representation, Presentation, Perception and Interpretation. If your long-term goal is to create interactive visualizations, I recommend you use an resource such as this to understand the 'theory'.

If you're looking for something more "quick and dirty" to get the job done once - that is, if you're less concerned about understanding the theory behind presenting data, and more interested in a one-time presentation - a few more details on your specific data would be useful.


I recommend the book of Alberto Cairo - "The functional Art". His website has a lot of resources and he is very active in social medias. I use his materials for lectures because they are easy understandable and cover the whole process from understanding visuals, encoding information for target audiences and understanding data.

Another good and profound source is the book "Information Visualization: Perception for Design" - Colin Ware This book digs deeper into how perception works and on advanced techniques to encode information for visualization.

There is no, let's say 10 step quick guide for interactive viz, because you will touch lots of challenges in terms of human perception and interaction design, understanding of data, its aggregations and finding/telling the story in your data,...

So these 2 book recommendations are just 2 out of many.


The design and how you present it all comes down to the story you are trying to tell with the data, or the questions people are trying to answer with the data. There are countless ways to do this - well, poorly, and dishonestly - and you could do worse than check out any of Edward Tufte's books - I'm a fan of "The Visual Display of Quantitative Information" and "Envisioning Information."

Another excellent book this is a bit more 'hands-on' and practical around designing for screen-based data visualizations is "Information Dashboard Design" by Stephen Few (along with his other book "Show me The Numbers").

But again: the most important thing is to know clearly what you want to get across with the data, the tasks the user is trying to accomplish with it, and then present them tools to do that in the best way possible.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.