Working with Tree Maps
New to Xcelsius 2008 is the Tree Map component. Tree Map components simultaneously use color and size to represent data pairs, such as median income level and employee turnover. A Tree Map component is a collection of non-overlapping colored tiles that completely fill up a large rectangle (see Figure 5.28).
Figure 5.28 Tree Map components display data by size and color and support drill down.
Each tile represents a row of data. Its size corresponds to the relative contribution of a specific measure, such as sales volume. The color of each rectangle can represent a different kind of measure, such as profitability.
A Tree Map component automatically arranges the tiles based on size and then by color or shading.
Tree maps are pretty, but unless you can easily connect them to data they use, their benefits are limited. In a world where there are lots of different kinds of data to examine, it would be nice to be able to choose datasets as easily as you can with the XY chart examples outlined a few pages ago.
Figure 5.29 shows the spreadsheet used to create the dashboard shown in Figure 5.28. The dataset in column C determines the tile size on the tree map. The dataset in column C is used to set the shading of colors for each of the tiles. When you start thinking about placing your data in two columns, one of which shows up as tile size and the other as tile color, the setup of a tree map becomes particularly easy to envision. The greater complexity comes about by shuttling data so that it is conveniently easy for a tree map to use.
Figure 5.29 The spreadsheet setup for a tree map.
Rather than reinvent the wheel, it makes sense to reuse spreadsheet designs already developed and vetted. You can use one of the spreadsheets already prepared in this chapter (refer to Figure 5.23) for the tree map.
In fact, the spreadsheet of Figure 5.25 was actually used to build this dashboard. Basically, the data was swapped, and a few formulas were tweaked. There is also a little extra work involved in drilling down to detailed information based on the selected tile.
Before we leave the topic of tree maps, I need to mention a few things about them:
- The hover text in a tree map typically consumes a fair amount of screen space. It can easily obscure other relevant data. For this reason, the drill down data is placed below the tree map and not to the right of it.
- When selecting colors for high and low values, try to stay in the same color family and vary the brightness.
- Each data series in a tree map consists of a pair of correlated datasets—one column for the size and the other for color. If you want to add a second series, place the data immediately to the right of the first data series.
- The tile area, and not the tile length or width, is proportional to its underlying data. If sales increased by a factor of 9, the relative length and width of the tile would increase by a factor of 3. This is both a good and bad thing. Because the total area for the whole tree map remains conserved, the other tile sizes get scaled down by a lesser amount. Small values don’t get diminished so quickly. It is also more difficult to interpret because we are used to linear proportionality, but in a tree map, tile size is proportional to the square root of its underlying data.
While a tree map may be pretty to look at, it doesn’t do anything that an XY chart doesn’t. Actually, an XY chart can be easier to interpret than a tree map. If you stop and think about it, the data points in an XY chart are, by definition, already sorted.