In my previous post last week I complained about the state of infovis blogging, arguing that there is not enough freely available quality material around for novices who want to learn, despite new websites and little business pop up every day. Especially, there is not much guidance for people who want to become visualization experts.
So I thought, what if one is attracted by visualization and wants to start? Is there anything on the web that helps taking at least the first few steps? So, I came up with the idea of writing a very small initial guide (recipe) to help people take these first steps.
(Of course, I know there’s never one single right path. Here is just what I think it’s best from my personal experience and knowledge. If you have different or additional ideas please let me know.)
My recipe is simple:
1) Study a Lot
5) Seek Discomfort
Study a Lot
Yes, I said “a lot”. This doesn’t come from day to night, but it’s fun! Unless you are in your teens or twenties and thinking about what to do in college or are thinking about going back to college to study computer science or related fields, the only way is to take the best material we currently have in the field and study. Here is my personal advice on the best sources to start learning the craft of data visualization, ordered by difficulty.
- Show Me the Numbers: Start from here. SMTN is the most gentle and yet deep introduction to data visualization I know. While it introduces only basic charts and tables I often realize how better I am in designing any kind of visualization thanks to it. So don’t underestimate its value. This is the foundation for any other kind of vis design (and I see so many crappy basic charts around that I cannot recommend this enough). It also provides basic knowledge on visual perception and its implication for design in a very concise manner, which is extremely useful.
- Readings in Information Visualization (First Chapter Only): If you want to understand what infovis is and how it differs from plain static data visualization, you have to start from here. That’s the foundation. The book is actually a collection of research papers with introductory chapters, but for this reason it is a bit outdated (yet many of these papes layed the foundation of the field). The first chapter is so good and dense of relevant information that I would print a small book out of it if possible. After several years I still find myself referring back to this chapter from time to time.
- The Visual Display of Quantitative Information: Edward Tufte is not an easy read. Well, actually it is not even a read given the images vs. text ratio. His books are so beautiful and full of great pictures that you learn something just by flipping through the pages. But TVDQI is not that glossy and, ironically, it’s the most informative. Follow my advice, if you want to read Tufte start from here and go over and over through his guidelines. His focus in minimalistic design and relentless removal of the superfluos has had a profound impact on my way of seeing visualization design. And judging from the noise I see around many people should go back to reading this hard-to-penetrate gem.
- Information Visualization: Perception for Design: If you digested these three books above, you are ready for some solid science. Colin Ware’s book is a must for visualization experts but don’t expect to learn any design tip and tricks here. The book is totally devoted to perceptual issues and I cannot stress more how important it is. Here you learn things like: how visual percpetion is higly contextual and why it counts in design, how color theory works and how to harness it for your purposes, how some visual features pop out more than others, etc. This is another book I refer back very often as well, but to the contrary of the others I still feel I’d better go back to it and study it entirely again because I know only an once of it.
Read these books, take your time, reflect on their advice and techniques and be ready to refer back to them often. You won’t regret having read them, I promise.
Stealing can be good if done with the right attitude and intent. When I say steal I mean you have to be an hunter for good visualizations and the tricks each one has. You have to be ready to absorb these trickes and make them yours. Remember: there is a huge differece between consuming information and absobing the tricks of the trade. A visualization expert is one who relentlessly seeks great examples to learn. But certainly you need some good resources. Few ideas:
- Papers from TOP conferences. Try to download some papers from past editions of the InfoVis conference (now part of VisWeek) (you might need an account for this, but normally they are accessible from universities and libraries). Where do you start from? Every year there is a best paper award nominated by a committee. You can bet this is one you have to read. Or try to see if a paper is highly cited (use google scholar for this), this is normally (not always) a good thing.
- Works from top people and labs. There are a good bunch of renown researchers, designer, consultants, labs, newspapers, etc. Maybe you already have your favorite. Start there and be sure to learn the tricks. An excellent example is the set of articles Stephen Few’s has on his website. He also has an examples web page with bad examples and their re-design. On the artistic/inspirational side you might want to look at Ben Fry’s projects page. I recetly aslo enjoyed a lot the work of Bang Wong If you have an academic taste, you might want to give a look to the visualization projects at the HCIL lab at the University of Maryland, endless sources for good interactive research-based tools. Finally, if you are a journalist or are interested in the use of data visualization to tell compelling stories, be sure not to miss the data graphics in the NY Times. But again, do your own research and find your idols.
One final notice. Be aware of what you find on the web. A beautiful visualization is not always a good one. You’d better develop an appreciation for criticism, which directly leads to my next point.
Part of the learning process comes from detecting bad examples and try to image how to fix them. The good news is that one of the most succesful visualization blogs does exactly that: be sure to visit Junk Charts to understand how the process works and try to do it by yourself.
What is really important is not to limit yourself to criticism. The best approach, and the hard part actually, is to redesign the visualization in a way that the flaws or limitations you find are corrected. As soon as you do it you will notice that it’s not easy. Normally, you have to find a tradeoff between competing needs.
Even if learning from books, teaching material and examples is important, more important is to practice. Visualization design is something you cannot learn if you don’t practice.
Again: you have to practice if you want to become an expert. Repeat with me: I have to practice to become an expert. Ok. But what are the steps if you don’t know anything about it yet?
- If you are a programmer: choose a language and an environment and stick with it until you are confident with your results. There are so many libraries and languages out there that this choice can really be overwehlming. Don’t overstress yourself om that. I think the choice of a language/environment doesn’t really matter as long as you produce good quality results. Find the one with less “friction” for you. If you want to make interactive stuff on the web the Flash technology seems to be the most advanced. Protovis has also be in the hype recently, but I don’t want to give specific advice here. Make your choice and stick with it, especially don’t think that the latest and coolest technology will make your visualization better. You can make great stuff regardless the technology you use.
- If you are not a programmer: you need charting tools that give you some freedom on the design side. The best choices I know are: Tableau, Excel, and R. Yes, I said Excel! If you don’t believe you can create great stuff with Excel check Jorge Camoes’ Excel Charts Blog and you’ll change your mind. Jorge’s blog is another proof that regardless the tool you use you can make great stuff.
Once you are set, the next step is to find interesting data. My suggestion is to find data that is interesting to YOU first. If you cannot amuse yourself it’s very hard you can please others. And then make sure you have interesting questions to ask. Again, coding a visualization for the sake of it is not very productive, unless you want to make a piece of art, and starting from some few questions is a key element of producing insightful visualizations.
Ok, you have the tools, the data, and the questions. Now what? You may decide to dig it directly and start producing your stuff. If you feel like that ok, but personally I always try to make some skecthes on paper or on a whiteboard first. There are pros and cons of this approach. The advantage is that you can probe your design more easily and faster and get a feeling for it. The disadvantage is that you can easily get stuck into it and fell in the trap of making it perfect on paper firts. No, the only way to really judge your visualization is to see it on your screen (or high quality print).
There is a huge difference between doing it for yourself and exposing your art to the world. This last step is painful and dreadful but maybe the most important one.
Today you no longer have excuses for not having your audience. The web is huge, maybe you already have twitter and facebook friends or whatever. You might even have a personal home page or, better, your personal blog. You decide what to use, but again you have no excuses for not exposing your art to the world.
There are three fundamental benefits you get out of this:
(1) You care more.
(2) You are forced to think about its use.
(3) You get feedback.
Plus, if you do it right you might become a star! Yes, think in this terms: I want to become a data visualization star. If you don’t even aim at that how can you become one?
Enjoy it … and let me know if you become a star!!!