Up to this point, my adventures in data visualization class have taken me down a rutted path, with lots of signposts and not a whole lot of original thought required. I’ve plugged provided code into consoles and patted my self on the back when they’ve behaved predictably (completely unashamed about feeling winner-ly) and used Inkscape to match the examples of others rather than coming up with my own. It’s been comfortable and I’ve had lots of opportunities to feel good about my abilities but at this point – but now – I am not so confident! I came into this knowing where my weaknesses were but not sure how they would feel when bumped up against, as I learn to systematically visualize data.
The most difficult part for me (beyond the annoyance of trying to match colors under what I now notice to be the extremely poor lighting conditions in my work areas — seriously! I’ve auditioned every light emitting device in the house for my workspace and none seem to do the trick) has been developing the heuristics necessary in order for the process of creating data visualizations to feel more natural. This has a lot to do with design but more to do with understanding visual encodings of statistical information. While I’m starting to sort of intuit what might be potentially interesting relationships between variables in a data set, especially after translating the data into something more or less like a narrative in my own head, or conducting a brief “hair on the back of my neck test” in order to determine what might be interesting trends or patterns to look for in the data, my processes are unsophisticated and I’m limited in what I can stumble on almost accidental ways. So, I wouldn’t say that my heuristics in regards to manipulating data into visualizations are non-existent, I think they are developing, but I need to expand my imagination into some new and unfamiliar territory.
Oddly enough, the thought that what I am learning to do involves no less than figuring out how to map information in a meaningful way to a 2D space using graphical semiotics precisely representing data, provides me with some comfort in this challenge. I’m not just making it hard, it actually is difficult. It is actually a more impressive feat to convince someone that a short distance on a screen or sheet of paper represents a construct like a specific length of time than it is to convince them that the letters C-A-N-O-E represent a type of water-borne vessel (or a nice brewpub in Victoria, B.C. with tasty beers I remember all being served gassed with nitro). Are words more believable than statistical charts, and if so, for what reason? Are words more recognizably human, therefore more easy to relate to? Anyway, I forgive myself for having a difficult time with this, and am going to try to enjoy the challenge without being too tough on myself.
So iterative design in the creation of data visualizations… which I interpret to mean, “you really won’t know whether or not you have turned down the right road until you get all the way, or almost, to the end, at which point you may or may not have time to backtrack and decide on a different road to venture down.” So it is. This suits me fine, in fact very well, as I am alright learning not just from, but by making mistakes.
Sketching on paper prior to doing any hard-core data visualizing is a great way to shorten trips down the wrong path, especially for folks like myself who are still fumbling with the means of sketching any other way. Here are some I created this week for a homework assignment. Pardon the miserable scan:
Data source for sketches: Sample dataset shipped with RStudio, “airPassengers”