Database Reference
In-Depth Information
Table 3.12
The table of cluster centers.
Cluster 1
Cluster 2
Cluster
N
Overall
population - all
clusters
...
Clustering field 1
Mean(1, 1)
Mean(1, 2)
Mean(1,
N
)
Mean(1)
Clustering field 2
Mean(2, 1)
Mean(2, 2)
Mean(2,
N
)
Mean(2)
.
.
.
.
.
Clustering field
M
Mean(
M
,1) Mean(
M
,2) Mean(
M
,
N
)
Mean(
M
)
In the general case of
M
clustering fields and
N
derived clusters, a table
summarizing the cluster centers has the form shown in Table 3.12.
Analysts should check each cluster individually and compare its means
for the input attributes to the overall population means, looking for significant
deviations from the ''typical'' behavior. Therefore, they should search for
clustering fields with relatively low or high mean values in a cluster.
2.
Comparison of clusters with respect to other key performance indi-
cators:
The profiling phase should also examine the clusters with respect to
''external'' fields not directly participating in cluster formation, including key
performance indicators (KPIs) and demographic information of interest. Nor-
mally, the cluster separation is not only limited to the clustering fields, but also
reflected in other attributes.
Therefore, data miners should also describe the clusters by using all the
important attributes, regardless of their involvement in the cluster building, to
fully portray the structure of each cluster and identify the features that best
characterize them.
This profiling typically involves examination of themean of each continuous
field of interest for each cluster. For categorical attributes, the procedure is
analogous, involving comparisons of frequencies and percentages. The scope
again is to uncover the cluster differentiation in terms of categorical fields.
This thorough examination of the clusters concludes with their labeling. All
clusters are assigned names that adequately summarize their distinctive character-
istics in a concise and simple way that will facilitate their subsequent use.
PROFILING THE CLUSTERS WITH IBM SPSS MODELER'S CLUSTER VIEWER
IBM SPSS Modeler offers a cluster profiling and evaluation tool that graphically
presents the structure of the revealed clusters. This tool, named the Cluster Viewer,
Search WWH ::
Custom Search