Bubble charts

Bubble charts are sometimes useful for visualizing data. This example uses color to encode country (mauve = Finland, peach = Israel, green = Italy) and size to encode number of patent documents. The labels identify USPTO art units. Overall, the visualization compares Finland, Israel and Italy in terms of the number of US patents which issued in 2012 to assignees located in those countries and which were allocated by the USPTO to one of five different art units. The five art units are:

• 2617 (cellular telephony)
• 2618 (radio/satellite communications)
• 2624 (image analysis)
• 2916 (a design patent art unit)
• 2913 (another design patent art unit)

You can easily see that, for Finland, art unit 2617 is the most significant one of the five. For Italy it’s art unit 2913 and for Israel it’s art unit 2624. In the underlying dataset, the Finland/2617 bubble corresponds to 127 patents, the Italy/2913 bubble corresponds to 54 patents and the Israel/2624 bubble corresponds to 48 patents.

For Finland, the next two most significant art units are 2916 and 2618 in that order, but you need to look closely to determine each bubble's size to get them in the right sequence. The Finland/2916 bubble corresponds to 51 patents and the Finland/2918 bubble corresponds to 46 patents. Difficulty in distinguishing bubble sizes is a downside of bubble charts.

For Israel, the next two most significant art units are 2617 and 2618 in that order, as is reasonably apparent from the bubbles’ respective sizes.

For Italy, the next two most significant art units are 2617 and 2624 in that order, but again you need to look closely to get them in the right order.  The Italy/2617 bubble corresponds to 23 patents and the Italy/2624 bubble corresponds to 18 patents.

The bubble size discrimination problem can be addressed by adding ranking values (e.g. 1, 2, 3...) to the bubbles within each color group, by applying different patterns corresponding to the number of patents represented by each bubble, etc. However, such techniques can distract the viewer without adequately addressing the problem.

Bubble charts are useful if you only want to see an approximation. But, if precision matters, bubble charts may not be the best choice. If you look back at my "Top technology sectors by country" post, you’ll see that I used data bars to compare Finland, Israel and Italy in a different context. Consider whether it’s easier to understand the data bar visualization or the bubble chart visualization.