Using scatter charts – part 1

ChartTypes-ScatterChartNowadays we see the rise of the scatter-charts. I begin to believe this chart is now more accepted than it used to be. One of the reasons is because of the general understanding people are getting for this chart. I think the scatter-chart is a very effective way to visualize multi-dimensional data over two or three measures.

Using this chart in QlikView is also very effective when you use it for example on dashboards. The main reasons for that are:

  1. They communicate lots of data in a tiny space
  2. They do the best job when determining the relevance of any ratio against its absolute outcome
  3. They give you an indication of distribution within 2 or 3 measures and thus showing relationships in your data
  4. You are instantly able to see outliers
  5. They are very powerful when it comes to information design
  6. And they will help you to zero in on the pieces of data that you are really interested in

This post will cover the basics of using scatter charts. We won’t bother you (yet) with the more powerful features that you could use when you want to deliver more insight in a more effective manner.

In the following example we chose to distribute the customer dimension over their revenue and the gross margin percentage that applies when you subtract the costs from the total revenue per customer.

Now when we add a scatter-chart to the QlikView Canvas, we start of with something like this:

scatter_init

Now in most cases this isn’t very readable and explanatory, yet. So what can we do to better communicate the data?

The next steps will help you to better design a scatter-chart:

  1. Hide the legend; this will only consume space as these items don’t really add much to the visualization, to much detail does not support for insight ‘at a glance’
  2. Setup the axes to have the labels along the axes, this will make more space for the bubbles on which we want the user to focus
  3. Strip off the caption and align the title of the chart to the left, this avoids redundancy of explanatory items. Also change the title to something more obvious for what you’re presenting your users
  4. Apply the correct number formatting, for example use percentages when presenting ratios (in this case the gross margin %)
  5. Adjust the font-size of the labels on the axes, we don’t want a user to focus on this first, this will be important when examining the chart in more detail. You might even want to consider to gray down the text on the axes.
  6. Use a flat styled bubble, this makes it easier for the user to see the differences between segments or individual bubbles
  7. Add target lines to the chart for spotting under- or over performance
  8. Place the measures on an axis that is more common to most users. In this case we present the revenue on the y-axis and the gross margin on the x-axis. Most users are used to reading absolute figures on the y-axis. So this would be more natural
  9. Color code the bubbles according to their performance which is also communicated by the target line. This will help the user to experience the design as more intuitive

After these adjustments the chart could look like this:

scatter_after

I think this is a lot better. What do you think? Any questions?

In a little while i will post some more powerful design and interaction features that make this chart more effective and fun to use.

Stay tuned…