30.12.15

Unbend Your Mind for Visual Analysis

Visual discovery with Tableau
Blank stares, then... Huh?

That's what I get when I say 'Visual analysis'.  People tend to think I'm talking about making a chart.  I am, but more than that, I'm talking about doing your data discovery visually with Tableau.

Chris Stolte gave a great talk on this a few years ago and I put it at the bottom of this blog as a Lunch n Learn.  I highly recommend you watch to see his brilliance in action.



Recently I was given a Tableau workbook to enhance/fix/re-jig that had over a dozen data sources and a dozen charts. Now a dozen charts is nothing for Tableau, yet the dashboards were taking forever to load and the filters were and weren't working in seemingly magical ways.

It took me a while to figure out the underlying logic of the workbook and then I realized that the analyst who had put it together was using the same mindset they had when the only tool they had for analysis and presentation was Excel.

The Excel logic goes like this:


Excel charts

1  We get a problem or request

2  We think of the chart we'll create in Excel

3  Then we go about getting the data into the table shape we need and perform the calculation(s) on the measure as needed

4  Then we make the chart


So the person who had built the Tableau dashboard had essentially created a bunch of pivot tables and put them on separate worksheets and then connected each of them to Tableau to make the charts.

This is not crazy - rather, this would be the most efficient way to do it in Excel world. Make a bunch of tables, then make the charts.

So, how to unbend the old Excel mindset?

FREE YOUR MIND AND THE REST WILL FOLLOW 

Connect to all the data just as you would to make a pivot table.

But don't try to make just one specific chart!


The Visual Analysis logic goes like this:


1  We get a problem or request

Tableau visual exploration
 Connect to the data

3  Explore the data VISUALLY.  Very quickly you can make a ton of charts - click through Show Me - add dimensions - play - see what the data has to say.  Discover what the data has to say in relation to your problem/question/goal.

 Then combine the charts on a dashboard to communicate the results.


Did you notice that big difference in these two processes?

Visual Analysis doesn't begin with the presumption of the outcome or answer.

Wrap your head around that for a moment.  I'll bet you didn't think that you were presuming the outcome with the Excel process.  But by reducing the data into a certain state for a predetermined presentation table or chart, you may have missed a lot of insight.

How often do you get asked by someone to give you a stacked bar or pie chart of xxx data?  That's because they have a hunch that that chart will show them the answer to a problem.  Sometimes it might.  But if you visually explore the data with Tableau, I'll bet you find a whole bunch of new insights.  Simply because it was never so quick or easy with other tools.  They weren't designed for this.

So watch Chris's video.  Give it a think.  And don't be afraid to go crazy with your data!