Click here to see the SAS code.
Click here to see the example.

Click here to see the contest scenario description.

---------------------------------------------------------------


DM Review Magazine Contest - Scenario 1:

Approach:

After experimenting with both grouped bars and line overlays,
I found grouped bars to provide the better visualization.

Axes:

Along the response axis, I convert the dollar-amounts to million dollars in order
to reduce the amount of ink/clutter of all the zeros.  I use 'Million U.S. $'
as the axis label to denote this.

I use 5 response axis tick marks, because that lends itself well to the 
range of the data values, and more ticks would be too cluttered.
At each tick mark, I turn the tick in towards the graph, and I extend 
light gray dashed reflines out from each tick mark (these reflines
are a subtle way to allow you to see whether the values have crossed
the tick mark values).

Along the midpoint/category axis, I use the 1-char month abbreviation rather 
than the full month names or 3-char abbreviations, or 1 & 2-digit month
numbers.  Anything longer than the 1-character month abbreviation
would have to be turned horizontal and a very small font used, to fit
under these dense/narrow bars.  I add a label "Month" to the right of
the axis, since people might not at first recognize the 1-char abbreviations,
and I also put "July-December" in the 2nd title at the top of the page
to reinforce the concept that the bars represent months.

Bars:

The bars themselves are colored by department.  Much care was taken in 
choosing colors that could be easily distinguished, while at the same time
not having too much garish contrast.  There is no spacing between the bars
within a group, but a small space between each month/group.  The bars are
subdivided into 2 segments (exempt and non-exempt), with a small gray dividing
line between the segments - to get this gray dividing line, I had to color
the outline of the entire bar gray (I experimented with several shades of gray,
to find one that was dark enough to contrast, while at the same time not 
being so dark that it overwhelmed the chart).

I ordered the groups of bars so that the group with the largest total
salary expenses were on the left, and the smallest on the right.
The color legend is sorted in the same order.  Within the group, of course, 
the bars are sorted by month (july-december).

I annotated a label at the top of each group of bars, with a summary/total
for all the values in that group of bars.  Also, the software I'm using allows 
web-based mouse-over charttips when the web version of the output is viewed,
so people can see the data values by hovering their mouse over the bar
segments, if one really needs to see the exact values.  An alternative would
be to look up the exact values in an accompanying table of the data values.

Legend:

Since there is adequate space inside the axes, I place the color legends there,
so it is close to the bars.  I also created a special legend showing what the 
two bar segments stand for (top segment = exempt salary, and bottom segment =
non-exempt salary).

Text:

All of the text and axis lines are black, and the background is white.
The reflines are a subtle light gray, so as not to stand out too much.
The "arial" truetype font is used, with anti-aliasing turned on to make
the edges of the fonts smooth (rather than blocky/jagged).

Observations/Results:

By far, the IT department has the highest Salary expenses of any department,
and this number is on the rise (whereas other departments are holding pretty
much steady).
The Marketing and HR departments have, by far, the smallest total salary expense.
Also, the IT department has had a dramatic increase in the salary of 'exempt' employees.
Also evident from the chart, both the Sales and Engineering departments are
mostly 'exempt' employees, whereas the Manufacturing and Ops are mostly 'non-exempt'.
And the Accounting department is about 50%/50% exempt and non-exempt.

Back to Samples Index