Data Viz Done Right

February 17, 2012

Pies and Cylinders: The uneducated continue to spread the virus

1 comment

Quite often I listen to presentations from other manufacturers on an “Insights & Analytics” share group.  Much of the content presented is directly applicable to my work; the education and ideas are great.  Unfortunately, far too often the presentations are of poor quality.  One thing I need to learn is to look past these “mistakes” and focus more on the story, but I’m struggling to do that.

Here are couple examples from a recent presentation:

image

These pie charts took several minutes for the presenter to explain, which is an immediate sign of their weaknesses.  The only part that works well is that the slices are ordered by the amount of time.  Other than that there are problems throughout.  I’ve gone into these in plenty of detail in the past.

The second charts that took my breath away are these 3D cylinder column charts.  I don’t see too many of these, for which I’m thankful.

image

There was no explanation whatsoever for how to interpret these charts.  My eyes take me back and for between the charts in an effort to compare like colors.  I can’t be convinced that was the intent though without the background commentary. 

I did a quick Google search for cylinder chart and the first site that came back was from anychart.com.  My terror increased when I read their definition.

Cylinder charts are column (or bar chart) that use cylinder shaped items to show data. Although cylinder charts do not add any additional data, sometimes using this shape allows to achieve a better visual appearance of your data.

The best visualization of Cylinder charts can be seen in 3D mode, so we will present all examples of them in this mode.

Seriously?  They achieve a better visual appearance?  How?  Their best us is in 3D mode?  OMG!  You really should checkout their site.  The examples are frightening!

So how can the cylinder charts presented be improved?

  1. Change the cylinder to plain old bars
  2. Add a scale
  3. Make it a clustered column chart.  This would allow you to compare the categories for each color to see the relationship between “Likelihood to Shop” and “Ease of Use”
  4. Consider a scatter plot

These are incredibly simple changes to make, which is what frustrates me so much.  It actually takes more work to make it look this bad versus using Excel’s defaults.

I struggle with charts like this each and every day, but the culture needs to change.  Have you been able to influence change?  How?

1 comment :

  1. Andy, these charts are amazing (and of course in this case, amazing = horrifying!). In terms of influencing change, in everything I do I try to emphasize simplicity and ease of interpretation over sexy glitz that adds no informative value. Over time, my hope is that both designers/analysts and audiences will see the benefits and demand better data viz. But I agree that it is a slow battle. Keep fighting the good fight!

    ReplyDelete