22.5.11

Chart Art

I recently read Stephen Few's blogpost on David McCandless and his infographics, which spawned a month of debate on FlowingData regarding the value of colorful artistic data visualizations and the importance of presenting data in ways that inform and align with principles of perception and cognition.  While beautiful, these types of visualizations don't help those of us working with analytics that are trying to dissuade management of being sold on gaudy, information-hiding,  expensive dashboard solutions.




I do enjoy the trend in using data to 'beautify' information in journalism, but I personally don't spend the time trying to find the meaning in most of them.  I prefer the simplified work from the likes of NY Times or the Guardian.  When I'm reading the news, I expect the graphic to quickly clarify information in the article, not require further analysis.  That's what analysts are asked to do for the companies we work for.  Tell the story. Quickly and simply.  If we expected management to do another level of analysis with the data, we'd go back to the days of giving them spreadsheets of tables or horrific Excel charts like the default pie chart above.

For fun, I tried to make a bad chart with Tableau. Tableau is the anti-Excel of analysis. With Excel I grab the data and make a chart. Then I spend 10 minutes un-doing all the chart junk. The reverse is true with Tableau; I throw the fields on the shelves and have to spend a lot of time trying to junk up a chart. A lot of time. Even then I couldn't make really bad chart art. I hate to admit it, but with Tableau, even a pie chart with 5 categories speaks fairly clearly.

As for the difference in marital status between Canadians and Iranians, well, for Iranians it seems it truly is 'Until Death Do Us Part'. Click on the pie slices or bars for view by age groups.


No comments:

Post a Comment