Data Visualization – Cool is Not a Key Driver!
At my job, we like to present data in nice colorful charts - often through a snappy dashboard solution. Top management in today's business seems to have a certain predilection for charts and graphics. Their love for charts and graphs is so endemic that the query and OLAP tools salespeople have learned the importance of underlining the charting capabilities in the tools they are trying to sell. No such demo is complete without creating a few charts. In reality, very few reports made by query and OLAP tools actually rely on charts. Tables with data placed in rows and columns still rule. However, charts do look a lot cooler.
Diagrams and graphics are definitely useful though, and the market for specialized data visualization tools is getting more mature. Data visualization tools have been around for quite some time, especially useful in trying to understand large amounts of data and its relationships. Until only a few years ago, data visualization was often classified as a form of data mining, given its complex usage. In order to distinguish data visualization from other data mining techniques, it began to be called visual data mining. Today it is most often simply called data visualization. These days, its appeal is broadening to a wider business audience instead of a select few.
How is data visualization adapting itself for a larger audience? Easy, it is getting back to basics. Forget the fancy three-dimensional hyperbolic tree structure charts or rotating three-dimensional scatter plot charts that can visualize seven different data columns by using size, colors, symbols, symbol inclination and the X, Y and Z axes. Many business users get confused already when more than four different things are shown in a chart. The most efficient data visualization tools that aim for a mass market focus on simple charts such as the pie chart or the bar chart. A single pie chart may not visualize more than two different data columns, but it can hardly be misunderstood.
In order to distribute data visualization to a large user community with different competencies and skills, commonly used charts should be employed. Just open a newspaper and look at the graphics they use when illustrating things (apart from photos that is). Most newspapers will stick to intuitive pie charts, bar charts, line charts and maps. This is what most readers will understand. When was the last time your local daily tried to illustrate an article with for example a cheerfully colored radar chart? And if they did, show it to your neighbors and friends and see if they are able to explain what they are seeing. Many people will not be able to easily understand information in that form.
Sure, whatever is colorful and rich in information may look cool and impressive at first glance. Colorful three-dimensional diagrams can look really advanced, but is it useful? Can the results be distributed to the intended audience without questions about what it all means? Compare with Rubik's cube which was cool too when it came in the late 1970s. A load of people fell for the craze. In the end though, most people never managed to solve this colorful three-dimensional challenge - unless they took the cube apart and rebuilt it with the colors put in the right place.
Data visualization is like most everything else: it can get as complicated as you want. However, visualizing data is no foolproof method in order to understand what the heck is going on. Depending on the competence of the users, the right presentations need to be elaborated. As uncool as it may be, simple pie charts and bar charts can still take you a long way in understanding and communicating what is going on.
For people with more specific needs, typically researchers and mathematicians, there are a plethora of different and very advanced graphical representations. (These are also the same people that often actually managed to solve Rubik's cube without taking it apart.) Just keep in mind that, as an example, a three-dimensional scatter plot may be completely incomprehensible to the uninitiated business users (and there are quite a lot of them out there). Any visualization will be of little use, unless it can be clearly communicated to the concerned users and allow them to act upon the results in the best possible manner. If not, data visualization will just look cool, but certainly not be a key driver for the business. And it will probably be a rather expensive way to look cool.One of the most unnecessary "cool" charts is probably the three-dimensional line chart. Such a graph will add absolutely nothing compared to the classical two-dimensional line chart. With the right colors, the three-dimensional line chart does look pretty fancy though. Given the absurdity in complicating something simple and perfectly useful, i.e., a two-dimensional line chart, I would say that in this case the right colors would be orange-blue, Mediterranean brick-green, watery cream and oxblood-yellow. This would be a chart that would rightly be surrealist even to top management, no matter how cool it may look.
For more information on related topics visit the following related portals...
Business Intelligence (BI) and
Gabriel Fuchs is a senior consultant with IBM. His column Reality IT takes an ironic look at what real-world IT solutions often look like - for better or for worse. The ideas and thoughts expressed in this column are based on Fuchs' own personal experience and imagination, and do not reflect the situation at IBM. He can be reached at firstname.lastname@example.org.