Graphics Reference
In-Depth Information
1.
1. State the goal.
Starting with the goal provides the frame of
reference as to why the audience should even be thinking about this
problem. For example,
“Identify strategies for microstores: an analysis
of complementary product categories to standalone.”
This can also
indicate the metrics of importance, the data used, the scope of the data,
and the analysis.
2.
2. Provide an overview.
The overall visual representation of a graph
may be unfamiliar and confusing to some viewers, with colored dots
and lines everywhere. A high-level indication should focus on the
following:
a. 2a. What are the nodes? An example might be, “
Nodes are products
we sell
.”
b. 2b. What are the edges? An example might be, “
Links indicate when
two products have sold together
.”
c. 2c. What do the size and color indicate? An example might be,
“
Nodes are sized and colored by revenue. The link width indicates
the number of times two products were purchased together
.”
d. 2d. Identify a few familiar recognizable data points. This moves the
story from the general concept into specific familiar information that
is relevant. An example might be, “
So, for example, this big purple
node represents light bulbs, our top-selling item. Or, this wide line
represents the connection between screwdrivers and screws
.”
3.
3. Provide a specific analysis.
Once the audience is familiar with
the objective, the structure, and a few data points, individual patterns
can be identified and explained. An example might be, “
These two
clusters to the upper-right corner represent lighting products and
paint products. Note that they are not strongly connected to our other
products—meaning that we could consider a microstore based purely
on this product segment.”
4.
4. Explain the anomalies.
The audience may be interested in what is
highly visible, such as outliers and anomalies (for example, the discrete
little cluster in
Figure 6-3
)
. These should not be ignored but explained
as part of the process to help familiarize the viewer with what may be
important or unimportant. If these anomalies are unimportant, it might
be useful to completely remove them if the presentation is brief or
include them and explain them away immediately after step 2d. An
