In the print world, we often use Comparative Charts to show trends in data or simply save space by showing 2 types of data simultaneously. These "overlayed" charts are a really handy way to quickly display a correlation between 2 distinct types of data without resorting to a side-by-side column or bar layout which doesn't make trends or percentages of data as clear as they can be for the audience.

You can whip up a chart in this style pretty easily in Keynote using the technique outlined below. As with any type of charting exercise, care should be given at each step of the process to ensure that you're accurately documenting and comparing your data. While it's beyond the scope of this quick tutorial to explain the concepts of data normalization and data-type congruity, it's important to bear in mind that this technique can be misused to artificially create apparent trends between data types where no such trend exists - use your head in determining if this technique is right for your data.

If you'd like more general information on creating accurate, compelling charts and data displays that covers the more abstract science that this technique approaches, we highly recommend Tufte's The Visual Display of Quantitative Information, widely considered the definitive work on charting and data visualization theory and practice.

The Test Case

Our test case will be a simple one - comparing total Sales by Region and Returns by Region for a product manufacturer. We'll end up with a Column chart displaying the sales, with an overlay Line chart outlining total product returns across the sales figures. For the screenshots, we'll use our Palo Alto theme and assume the entire presentation is being produced with this theme, although the technique can be used with nearly any theme you like.

To begin, we'll incorporate our data into 2 different charts. In this case, we're using a unique slide for each to start out with, using the default Column Chart style for each. Then, we create a third slide and copy & paste the charts from slide 1 and slide 2 respectively. (figure 1)

 

Figure 1. We start with unique charts for each set of data - in this case Sales by Region and Returns by Region. On a new slide, paste each of these against each other to start.

 

 

Figure 2. Normalize the data by setting the high and low values displayed on the Y-axis to the same values - by creating a 3rd "Region" which we'll hide.

 

The first thing we'll notice is that the data visualizations match each other, in terms of size, very closely. This is because we do not have consistent range values between the two data sets - the rate of returns is actually much lower than the rate of sales, and so the chart for the Returns by Region data tops out at a lower Y-axis value, though Keynote will draw the charts at the same relative size. As our low-end value - zero - is the same between the two data sets, we'll need to normalize against the highest value displayed. The Sales by Region chart, in this case, is showing "100" as the highest number on the Y-axis, so we'll want to normalize the Returns data against 100 to shift the data visualization accordingly. We'll do this by adding a baseline row (essentially a "region 3") to the Returns data (we'll hide this one later), setting the value for each year (2001-2004) to 100. The best way to compare these types of data is by Region, so we'll change from a column-based display to row-based on both charts - so the legend for each should now show Region 1 and Region 2, as well as Region 3 on the second chart. Now that we've normalized the data, we can turn off the legend and X & Y axis Category and Value labels respectively and simply work with the charts themselves.

To continue, we'll change the Returns by Region chart over to a Line Chart style, which will overlay nicely against the Column Chart of sales data. You'll notice a straight line for the 3rd Region, or our baseline/normalization data, running across the top of the chart. We'll hide this data by editing the line style for this line item by itself in the inspector. First, select the line chart by clicking on one of the lines. Then, click on the top line again to select it by itself - you should see highlighting elements surrounding the line. First, in the Graphic Inspector, we'll adjust the stroke value to 1 pixel (you can't go any lower), and modify the stroke color to a lighter version of the stroke the line will be laying against (and in this case, also modify it to a dashed line type) - or to the color of the stroke itself if you're lining up against a solid line. Now, switch over to the Chart Inspector, select the "Series" tab, and look for the Data Point Symbol selector, which you'll set to "None." (figure 3) We've effectively hidden our normalization data at this point, and can proceed with customizing the appearance for more clarity.

 

Figure 3. Once we shift to a Line Chart style for the Return data, we’ll hide the normalization data (Region 3) by editing the line style of that line in the chart by itself.

 

All that's left is to make the Line Chart stand out a bit more against the Column chart by changing the colors used for the lines, and adding the finishing touches. As we did with the line of normalization data, we'll select each line on the chart and modify the stroke color to a pair of contrasting colors. We also like to drop the default stroke width down a few points so that this overlay is accenting the Column data - in our case we're dropping to a 3 pixel stroke, down from the default 6. Make sure the charts are aligned if you haven't done so already, and then scale the line chart via the selectors on the sides to line up the points on the lines with the outside of the columns. If you'd like to, you can also hide the Y-axis gridlines altogether at this point by selecting the line chart, opening the Chart Inspector and changing to the Axis tab, then deselecting "show gridlines" from the y-Axis selector - though this is by no means a necessary step. You can help your users to identify the data by adding additional legend colors as well, by adding small square shapes, changing the fills to the correct colors used for each region, and then adding those blocks alongside the existing legend markers. A finished chart is shown in figure 4.

 

Figure 4. Set the Line chart stroke colors to contrast the existing chart palette used in the Column chart for better clarity, and add finishing touches to wrap up the combined chart.

 

As you'll see in the finished chart, we now have a fully combined single-page chart displaying both sales and return data clearly and cleanly. As we've normalized the data, we can be sure that we're showing an accurate trend between sales and returns by region - in this case a slightly diminishing rate of returns in line with growth in unit sales. This can be a highly effective technique for data visualization, as long as you're careful to combine data that can display accurate relationships and normalize their respective data sets to ensure that trends are not being over (or under) reported.