The goal of visual data analysis is to make obvious implicit and otherwise difficult-to-discern relationships. I started out this chapter by saying that every dashboard should have a story to tell. Sometimes, it’s the dashboard designer who knows exactly what has to be said and is in need of a way to masterfully present the message. Other times, the dashboard lets the data speak for itself, by making it easy for the end user to examine and explore the data with ease and turn over stones that would otherwise be left untouched. I refer to the latter as “planned serendipity.”
This chapter presents key issues, possible strategies, useful techniques, and hidden gotchas that tend to come up when presenting data visually. Along the way, many of the principles and techniques are shown in action.
The best place to begin is to ask, which components do I use? Then you can tackle, how do I tame the data? One answer is to put the data on a timeline. In this manner, you can see trends but not be overwhelmed by a dizzying array of distracting information competing for your attention.
In some cases a dashboard may be otherwise well designed, but the cosmetics get in the way. Xcelsius automatically enables data animation, which gives the dashboard a certain coolness and is designed to “wow” the audience. Unfortunately, when you are trying to analyze patterns and trends, the jittery behavior of this feature can be downright distracting!
As you get more sophisticated in your dashboard skills, you are bound to combine several components so they work as one. People often forget to make components visually blend together as if they are one larger component. Sometimes all it takes to glue them together is a single visual background. Sometimes it makes sense to stack data together so you can see all the data together at one time. Stacked data and its components become further empowered when you can drill down to get at the underlying details.
A common perceived limitation of pie charts is that they are only well suited to situations in which the various slices of the pie are roughly similar in size, and there are not too many of them. However, by dynamically grouping the smallest slices, you can use pie charts in many other situations as well.
Many of the Xcelsius charting components, such as column charts, are designed to handle histogram-like data where one of the axes is continuous, and the other varies in discrete measures or categories. There are times when it is necessary to get more quantitative and display two or more measures. This is where XY charts, bubble charts, and tree maps come into play. Standing behind these charts can be a variety of different kinds of datasets, waiting to be visually mixed and matched. I introduce a technique of using correlated list boxes to seamlessly select the datasets to be displayed. This technique takes context switching to an extreme.
In this chapter, you saw a solution for rendering bubble charts when the bubble sizes have negative values. You also learned how to set up tree maps. In addition, you learned about chart scaling because it is important to be able to fully control a dashboard’s visual elements.
The theme of visual elements continues into Chapter 6, “Single Value Components: Dials, Gauges, Speedometers, and the Like.”