Home > Articles > Home & Office Computing

  • Print
  • + Share This
This chapter is from the book

Building on Your Visualizations

The Spinner component is not the only component that’s suitable for setting a point in a timeline. You could instead use Slider, Dial, or Calendar components, to name a few.

The data becomes more interesting when it is not set in a vacuum. You might, for instance, want to compare the number of unique daily visitors to a website to the number of page views. To do this, you would need to add a second data series to the chart. Because comparing unique visitors and page views is really like comparing apples and oranges, a column or bar chart is not suited for this task, even if they were both plotted over the same range of dates. Line charts and combination charts work better for this purpose. As long as you don’t need to make use of the Xcelsius Alerts feature, the Combination Chart component is the best choice in this situation.

Putting Your Data onto a Timeline

Figure 5.6 shows how a combination chart can be used to present two data series: The vertical bars represent visitor count, and the line graph represents page views. Because visitor count and page views are not exactly the same kind of quantity, you need to make use of a dual-axis facility.

Figure 5.6

Figure 5.6 A combination chart is well suited for simultaneously displaying different kinds of information along a common axis.

The following are important design features of this combination chart:

  • You can make the chart title and/or subtitle dependent on the underlying spreadsheet content. In this example, the subtitle is pegged to cell C5, which changes every time the day number in the Spinner control is changed.
  • The Spinner control title and rectangular background are purposely similar in appearance to the Combination Chart legend. This allows the dashboard user to perceive the Spinner control as an actual part of the combination chart.
  • The plot area of the chart is a distinctly different color or shading than the area immediately behind the chart. This helps the visual data to stand out. The horizontal gridlines are visible, but they don’t compete for attention with the chart data. In particular, only the major gridlines are enabled. If minor gridlines were enabled, the chart might be a little too busy.
  • The labels along the axes and in the legend appear in boldface, making the chart easier to read. Using contrasting colors or shades between the chart labels and their background also helps the readability.

There are some hidden wrinkles that you need to be aware of related to combination charts. Figure 5.6 shows one of them. The primary axis ranges from a value of 1000 (a nonzero number) to 6000. The secondary axis ranges from a value of 0 to 120K. As you cycle through the days, as shown in Figure 5.7, notice that the scaling is not exactly proportional.

Figure 5.7

Figure 5.7 The scaling in this chart is not always proportional.

On day 147, the maximum value of both axes jumps up 50% (from 6000 to 9000 and from 120K to 180K), but the minimum values do not change uniformly. The primary axis originally starts at 1000, and on day 147, it grows to 3000. The secondary axis originally starts at 0, and it remains unchanged when the timeline advances to day 147. Clearly, the scales do not remain proportional as you advance the timeline.

You can force these scales to be proportional, but to do so, you must have complete control over the scaling, and you may not always be happy with the chart appearance. You can experiment with the file ch05_DataViewer.xlf, which provides a solution.

You may need to be aware of a couple other things. Dual-axis charts are generally supported in Xcelsius 2008. If you plan on displaying three or more data series in a chart, at least two of the series will have to share either the primary axis or the secondary axis. If your data series contains similarly valued items (such as percentage of efficiency or market penetration), this would not be a problem. If the values between data series vary significantly, this could be problematic. Consider the example of unique visitor counts and total page views. If you want to plot the ratio of page views per visitor, you might find numbers typically varying between 10 and 25. When you try including these as an additional data series in the combination chart, the data becomes flatlined, as the numbers are too small for either of the primary or secondary scales. To cope with this issue, you have several strategies available.

  • You could put the page views on the same axis as the visitors and place the page views per visitor on the other axis. Unless the data series sharing a common axis have similar values, this is not going to be a very effective solution. In this particular case, the page views dominate. The visitor count is visible but too small, resulting in loss of meaningful information.

    A common technique for dealing with quantities that are vastly different in order of magnitude is to apply logarithmic scaling instead of linear scaling.

  • You could apply context switching so that only one data series is displayed at any time, but the user would have complete freedom to choose which two data series you want to view.
  • You could overlay a line chart on top of the combination chart. The line chart would need to be precisely positioned. Its background would have to be disabled so it is fully transparent. You would not display the line chart axis labels. The line chart axes could be hidden as well.
  • Instead of overlaying a chart, you could make a separate chart that is pegged to the same timeline as the main chart. If you are going to follow this strategy, and the timeline shifts the displayed data to the left or right, you should place the separate chart directly below or above the main chart, not to its left or right (see Figure 5.8).

    Figure 5.8

    Figure 5.8 A possible layout for two charts on the same timeline.

  • + Share This
  • 🔖 Save To Your Account