On Sparktweets, Sparklines, and Tiny Data

25May11

Recently, the hashtag #sparktweets made the rounds on Twitter for a bit, before it faded into, well, the dull background roar of 140-word mini-musings that is Twitter. The original post about them is here. They’re a clever way to present visual information using standard ASCII characters inside Tweets – and other short, space limited media. They’re named after Tufte’s “sparklines“, and the original article about them can be found here: http://zachseward.com/sparktweets/

But what are they? And are they useful?

First, examples, from my own Twitter feed (@epigrad).

This is a sparktweet of the # of WHO confirmed pandemic H1N1 flu cases in the U.S. in 2009: ▁▁▁▁▃▃▂▂▃▇▃▁ . Neat, innit? Or maybe not. There’s clearly some problems. There’s no x-axis (the blocks are months, btw). There’s more importantly, no Y-axis, no scale. In this case, one can easily figure out that the flat line at the bottom is 0, since the pandemic didn’t really kick off until spring of 2009. But many figures, those who don’t start at 0, like the popular WSJ.com sparktweet on unemployment, can end up looking hugely deceptive, suggesting things like the unemployment rate tripled in a month period. That’s…bad.

They’re also pretty low resolution – they can only show 6 unique values, so a single large spike may drown out others.

So, in one author’s opinion, are they useful? Well – maybe. They’re certainly neat, and a clever use of ASCII characters. They’re lower resolution than a proper sparkline, but then Tweets are fairly low resolution to begin with. It’s not a medium meant to convey rich information. And while we’re on the topic, its not as if the original concept behind the sparkline is without its faults. Tufte is a brilliant mind in data visualization, but I’ve found them to be unintuitive to someone who doesn’t know what they are (I’ve always been of the belief that visualization should make science more accessible, not less) and only really allow broad-scope analysis of a time series – is it going up, is it going down, is it noisy, is it cyclical?

So yes, they can be useful. If you want to convey a simple time-series or bar chart of data, preferably that has a 0 origin already, and what you’re trying to show is a clear, linear (versus exponential or the like) trend, then a sparktweet just might do it in your feed.

Or you can just make a proper graph and link to it. But what fun is that?

Curious? Make your own: http://www.datacollective.org/sparkblocks.html

Advertisements


No Responses Yet to “On Sparktweets, Sparklines, and Tiny Data”

  1. Leave a Comment

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


%d bloggers like this: