As sophisticated as business intelligence tools have become over the years, when it comes time to actually present the intelligence or information as a report or in summary form, most of the time you'll still find yourself looking at traditional graphical tables (pie charts, line graphs and bar charts). Maybe they'll be drawn creatively (e.g., USA Today's uncanny ability to turn ordinary bar charts into pictures of things such as piggy banks, toasters and airplane engines) – but they're still basically bar charts.

The problem with this sort of tabular presentation of information is not just that it lacks pizzazz. The problem is actually that some kinds of information are so rich and complex that they cannot be captured effectively in two-dimensional ways. Some information cries out for a more distinctive visual treatment, one that presents information more powerfully.

The good news is that the potential certainly exists to deliver information in richer, more complex visual forms. Advancements in this field – called "data visualization" – have been enabled by today's powerful computers with GUI capabilities that can do the heavy lifting required to present information with more clarity and more power. The bad news is that too few technologists have the combination of technical and communication skills necessary to create the most effective statistical graphics.

Effective Data Visualization: Capturing Both Detail and Context

The advantage of advanced data visualization techniques over traditional charts is their ability to drill down on the data while retaining the context of the entire domain. For example, one of the early uses of data visualization we completed at Accenture was a kiosk display for an internal meeting that presented a digitized globe on which countries rose to varying heights depending on the amount of business growth in that country. Further, the countries appeared to rise on striated pillars that were color-coded to reflect the industries in which the business occurred. With this presentation of information, you could see specific information for each country. At the same time, using the entire globe meant that the larger context was also preserved. If the same data had been presented through more traditional charts, examination of one country would have meant that data for all other countries would have been eliminated from the presentation.

Advanced data visualization techniques also present large amounts of complex data in more digestible ways. One of the most effective data visualizations ever created is Charles Joseph Minard's map of Napoleon's assault on Russia in 1812-1813 (see Figure 1). The chart is six-dimensional, presenting time, location in two dimensions, direction of movement, troop strength and climate all in a single effective chart. Unfortunately, not all attempts to present statistical data are as successful, and many of them obscure more than they clarify. As with the chart showing Napoleon's assault, a certain degree of creativity is required. I hope I won't offend too many readers when I say that visual creativity is not necessarily the number one capability of those in our field.

Figure 1: Minard's Map of Napoleon's Assault on Russia

Learning from the Best

Fortunately, there are some guidelines based on experience that can direct our creative efforts. Using Edward Tufte's seminal work, The Visual Display of Quantitative Information, we can discern the following general points about how to create effective statistical graphics.1

  • Effectiveness: Don't hide information. We have all experienced the frustration of overlapping windows and HTML subpages that make it extremely difficult to find what we need. To be effective, information needs to be visible.
  • Accuracy: Visual proportions should reflect actual proportions. Financial line graphs, for example, are frequently guilty of presenting information in a misleading way. For example, consider a presentation about stock growth (we're optimists here) showing two separate stocks ­ one growing from $10 to $12, the other from $100 to $102. If you drew a graphic just showing those ranges, you might create the impression that both stocks had the same growth. Of course, they didn't. Shown with the proper context ­ a scale starting at $0 ­ the data visualization would correctly demonstrate that one stock had a healthy growth of 20 percent while the other grew just 2 percent.
  • Efficiency: Minimize your ink-data ratio and chart junk. Graphics that present dense imagery and supplemental details not relevant to the main messages create more confusion than clarity. Avoid heavy lines, excessive text or too many icons. Do not use three-dimensional charts to present two-dimensional data (a common flaw in today's graphics-loving culture). Less is definitely more.
  • Aesthetics: The best statistical graphics appeal to our senses, not just our minds. Yes, there is some subjectivity here, but the best charts have subtle colors, clean lines and a good sense of proportion.

You know the old maxim: A picture is worth a thousand words. The other side of that maxim is that a bad picture requires an extra two thousand words to explain it. As business intelligence applications strive to incorporate more advanced data visualization techniques, we all need to educate ourselves about the elements of good visualization and incorporate them into our work.

1. Tufte, Edward R. The Visual Display of Quantitative Information. Chesire: Graphics Press, 1983.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access