Databases Reference
In-Depth Information
The Scatter Plot viewer automatically adjusts the plot type depending on
the type of attributes selected and the number of observations in the dataset.
You will see more plot types in later tutorials as you explore additional
datasets.
Using the Control Center, close the correlation matrix and the scatter plot,
by clicking on the small red “X” in the upper right corner of each viewer.
Exercise 2.3
Use the VisMiner correlation matrix, histogram and scatter plot to answer the
questions below with respect to the OliveOil dataset.
The correlation matrix reports an inverse correlation between eicosenoic and
oleic acids (
0.424). Does it fully reflect the actual relationship between these
two acids? Explain your answer using snapshots of any scatter plots that you
used to answer this question.
a. Evaluate the distribution of stearic, eicosenoic, and palmitic acid measure-
ments using the histogram viewer. Do they appear close to normal, skewed,
or multimodal?
b. Develop two rules (guidelines) that could be used to classify observations
into regions. For example, a rule might be: If x
>
50, then assign to
category A. Be sure to include at least one rule that differentiates between
the North and Sardinia regions. Include selected scatter plots to support
your rules.
The parallel coordinate plot
The parallel coordinate plot (PCP) is the most powerful of the available
visualizations in VisMiner and due to that power can take more time to
interpret. Not all of the patterns are preattentive.
To open, drag the Iris dataset up to the display and release, then select
“Parallel Plot” in the context menu (Figure 2.16).
One of the strengths of the PCP is that it is not limited in the number of
dimensions that it can concurrently plot. As the name implies, each dimension is
plotted along axes that are laid out in parallel rather than orthogonal axes like a
scatter plot. Each observation is represented as a sequence of line segments
connecting points on each of the axes. To help you understand the PCP, let's
focus on just one observation. There are small red triangles at the top and bottom
of each of the axes. The triangles function as range end-points or filters. Only
