Data Visualization Tips: I Love Bubble Charts

Porter Thorndike's picture
 By | November 04, 2014
in WebFOCUS, Advanced Analytics, Advanced Visualization, Analytic Dashboards, Best Practices, Big Data, Business Intelligence, Dashboards, Data Discovery, data visualization, Data Visualization
November 04, 2014

I love bubble charts.  They're pretty and communicate a lot of data, kind of like me.  Any time I get a new data set, the bubble chart is my go to visualization to find something interesting in the data.  Why is the bubble chart so good at finding patterns in data?

The Bubble Chart handles large numbers of data points with eloquence.  Whether it is 5 data points or 5,000 data points the result is always something insightful.  Are the points spread out or clustered tightly together?  Is there an outlier?  I always hold my breath in the moments before that first bubble chart is rendered wondering what I will find.

Like its cousin the scatterplot, the bubble chart will plot points using the vertical and horizontal placement of two different metrics. The use of color is your third way to spot patterns in the data.  In analyzing data of every flight delay in 2013 I found no obvious patterns or outliers when plotting every plane by origin delay, departure delay and cancellations.  As soon as I added a color by of carrier the screen popped. There was a hotspot of a single carrier with very poor delay and cancellation averages.  Lastly, the bubble chart uses size to communicate YET ANOTHER METRIC.  It is this fourth opportunity to spot outliers that makes it the darling of my data discovery laboratory. 

One of the reasons InfoDiscovery is so good at leveraging visualizations like bubble charts is how quickly I can build them and how rapidly I can evaluate many metrics. Any time I get a new dataset I splash up a bubble chart in under 60 seconds.  What follows in the minutes after are a rapid fire evaluation of several different metric combinations of vertical position, horizontal position and size to find the patterns in the data.  Then the rapid evaluation of several different color by dimensions.  Lastly, I always know that I can use the InfoDiscovery drilldown hierarchy to jump up or down any level of my dimensional hierarchy.  I also know that at any time I can lasso, exclude and zoom in on a smaller subset of my data.

For this Analyst, the Bubble Chart and InfoDiscovery were love at first sight.

To learn more about the visualizations I love, and to view a live 30-minute demo of InfoDiscovery’s capabilities, please join me for a Daytime Demo running through Nov 12.

Oh, and you can read about my fondness for bubble-wrap vests and more here.