Figure 4: Cluster Density
b.
In the
Cluster Representations
pane, select Feature Distribution.

MIS688: Predictive Analytics
‐
6
‐
i.
The graph lets you compare the distribution of the variable in a particular cluster
against the entire dataset. You can change the Measure being displayed and the
cluster number in the Data panel on the right side.
Figure 5: Feature Distribution
c.
In the
Cluster Representations
pane, select Cluster Center Representation.
i.
You see a radar chart of the cluster centers (radar axes are the variables); you can
change the cluster number in the Data panel
Figure 6: Radar Chart of Clusters
c.
In the
Cluster Representations
pane, select Parallel Coordinate Chart.
i.
The axes are all normalized. Parallel lines between the axes imply a positive
relationship between the two dimensions. Intersecting lines imply a negative
relationship.

MIS688: Predictive Analytics
‐
7
‐
Figure 7: Parallel Coordinates Chart
d.
In the
Cluster Representations
pane, select Scatter Matrix Charts.
i.
You see the scatter charts of store clusters plotted between various pairs of
dimensions
Figure 8: Scatter Plots
27.
The fitted and forecast results are stored in the CSV file. You can open the saved csv file
and explore the three clusters that have been generated.
28.
From the
File
menu, select
Save
.
29.
Enter a name for the document.
30.
Choose
Save