Stack Graph Guide
When to use a stack graph
A stack graph is a classic method for visualizing change in a set of items, where the sum of the values is as important as the individual items. A stack graph is excellent for looking at revenue over time across several products, for example. Because a stack graph uses areas to convey numbers, they don't work for negative values. And in some situations it might not make sense to add up different data series (say, prices of different stocks over time). For these cases you may want to use a line graph.
Many Eyes includes two different stack graphs. The plain stack graph, described here, is appropriate for up to a few dozen items. If you have a larger number of items that are organized into categories and subcategories, you should consider the Stack Graph for Categories.
How our stack graph works
To see the exact value of a stripe on the graph, move your mouse over it. If you are graphing more than one data series, use the clickable legend to pick which graph to see. If you want to see more than one series at once, use control-clicks on the legend.
To highlight a point on the graph, just click it. A prominent orange "I" shape will appear, showing the width of the stripe at a particular x value. To remove the highlight, just click again. To highlight more than one point, use control-clicks. Your highlights will be saved with any comments you make, so you can easily refer to particular parts of the graph.
The "percentage" checkbox will adjust the graph so that each ribbon shows the percentage of the total for that point in time, rather than an absolute number. This option is helpful for emphasizing the relative sizes of the ribbons.
Data requirements
A stack graph takes a table where each column corresponds one data series. One special column corresponds to the x-axis labels. For example, in this table the "Year" column would be used for the x-axis labels, and the two other columns would define two data series to be graphed. Note that this example is natural for a stack graph since the values are positive, and when you add them up you get something meaningful: the price of a (not so healthful) breakfast.
A sample data table in the stack graph format is:
| Year | Price of Doughnut ($) | Price of Coffee ($) |
|---|---|---|
| 2005 | .75 | 1.42 |
| 2006 | .78 | 2.52 |
| 2007 | .81 | 2.15 |
A complication is that in some data sets the rows and columns are flipped, like this:
| Year | 2005 | 2006 | 2007 | |
|---|---|---|---|---|
| Price of Doughnut ($) | .75 | .78 | .81 | |
| Price of Coffee ($) | 1.42 | 2.52 | 2.15 |
In this case, choose the "Flip rows and columns" option when you first configure the graph.
Caution: Because negative values don't make sense for this type of graph, they will be replaced by zeros.
Expert notes
Stack graphs are a standard, useful chart. They can convey multiple levels of meaning at once, but have some drawbacks. It can be hard to judge the exact widths of a stripe in a stack graph, or compare the widths of two stripes. If exactitude or comparisons are of primary importance, consider a line graph instead. Another consideration is that some time series don't make sense to add. Adding the stock price of GE to the stock price of Intel is pointless.