VizWiz

Launch, grow, and unlock your career in data

January 29, 2013

Alberto Cairo: Three steps to become a visualization/infographics designer (a Tableau version)

11 comments

Two weeks ago, we had the huge honor of hosting Alberto Cairo at Facebook for our monthly Data Visualization Meet-up.  And wow was his presentation incredible!  People are still buzzing about it. 

Alberto did a wonderful job of taking us through the decision making processes he’s gone through during his years of creating visualizations and infographics for many media organizations.  One in particular that caught my attention was how he came about creating the cover for his book “the functional art”. 

Alberto walked us through the iterative approach he took to end up with the slope graph on the cover.  He sent me the data (which you can download here) so that I could emulate the design process in Tableau.  Using the “Next” buttons, you can navigate through his approach to see the story unfold.

11 comments :

  1. Nice job --- I've seen Cairos' viz before, but your interactive version of it made it "click". Also like your use of the "Next" button. ...

    ReplyDelete
  2. Nice stuff Andy. Slopegraphs in Tableau are almost amazing. Right now, I think they don't work because the label cannot be easily placed directly next to the line. Alberto's cover chart has the State abbreviation right against the line, so your eyes aren't drawn away to identify the line. In Tableau we need to use a legend still as the labels overlap.

    That said - your implementation is really nice. It looks gorgeous and is as good a slope chart as we could do right now. The Highlighting makes life much better**

    Is a Slope chart better than a scatter plot? Not necessarily. The slope chart makes DC stand out instantly thought.

    ** in v8 we'll do much more Browser rendering. This means you can use Hover actions in Tableau Public much more effectively as the Hover will be instant - this is going to make life a lot easier for charts like this!

    ReplyDelete
  3. Nice example! Yet one thing I noticed and same as I experienced while using Tableau - both professional and free versions, is, the limitation of customization. Among all the visuals, I like the scatter plot most, since it is straight forward and deliver the comparison between the higher educated % and obese % of the population. Yet the labeling is an issue - there is no label of NY! And it will be better if you can apply Color by State, which will display different state in different color. In that case, even if i cannot locate New York by label, I can still find it by color legend. Secondly, may be a good idea to add more details to the data point - say when mouse over, it shows the total population of the State. Third,I am not sure if you have normalized the data before plotting into the visuals. Since population in different states varies vastly, one shall take that into consideration and normalize the data before applying, using index instead of actual percentage of population will be more accurate. For the third point, I don't think Tableau can handle it or maybe just I don't know how.

    ReplyDelete
    Replies
    1. Nomad,

      What examples do you have as limitations of customization? I can help determine if they're indeed limitations only if you provide examples.

      Tableau intentionally leaves labels off that will overlap so that the visualization is not too cluttered. Can I force the marks to appear? Yes, but it would look horrible. The whole idea of a scatter plot is to find outliers which this does well.

      Coloring each date by the state would be a horrible design decision. There would be way too many colors, which would overpower the integrity of the chart. Again the purpose is to highlight outliers.

      I did not include population in the data set. If you re-read the beginning of the post, I clearly state that this was done to mimic the creation of Alberto's book cover, exactly as he showed it was done to us. And yes, Tableau can handle the indexing, it's only a matter of creating a calculated field, which takes about two seconds.

      Delete
  4. Andy, I'm impressed with your slopegraph in Tableau. I'm in Alberto's MOOC class and am trying to replicate it for an assignment. I've created the slopegraph but so far haven't figured out how to make it so that clicking on a geography in "States" will highlight that state's line. "States" doesn't appear to be a filter. Also, how did you make the lines gray before clicking? I've been studying your worksheets but haven't figured it out yet.

    Thanks!

    Jeff

    ReplyDelete
    Replies
    1. Hey Jeff. I'd be more than happy to hop on a webex with you if that works for you. There's a link to my email address on the upper-right of the blog.

      Delete
    2. Thanks, Andy. I sent a message through the form on your blog.
      Jeff

      Delete
  5. Interesting graphic interface, but what is the story you are trying to tell? Are you illustrating a causal relationship between obesity and lack of education: the old stereotype that overweight people are uneducated? Groupings by states like this can be particularly misleading. The graphic tells a story, but is it one that is valid statistically? Beautiful graphics can be used to inform and to mislead.

    ReplyDelete
    Replies
    1. The purpose of this post was to demonstrate the iterative process that Alberto goes through and to determine if I could reproduce in Tableau what he did in Illustrator. The question about the causality would be better directed at him. I'm sure he'd be happy to respond if you reached out to him on Twitter.

      Delete
    2. I think all it shows is a limited inverse correlation between obesity and education. It's all within a small range and says nothing about causation. Its the kind of misleading graphic politicians and other ne'er do wells like to use to push a dubious political or social point.

      But as a process it is interesting.

      Delete