Data Mining Techniques for Segmentation - Data Mining Techniques in CRM: Inside Customer Segmentation - page 107

Table 3.13 The means of original attributes of particular importance for each cluster.

Clusters

Cluster Cluster Cluster Cluster Cluster Cluster Total
1 2 3 4 5 6

VOICE CALLS 345.3 329.1 105.5 133.8 136.7 107.4 165.6

SMS CALLS 292.9 44.9 128.1 42.3 57.8 7.8 55.4

VOICE ROAM-
ING CALLS

104.5 1.3 0.8 25.7 1.2 0.4 3.5

MMS CALLS 8.4 0.5 0.3 0.2 4.1 0.1 0.7

GPRSTRAFFIC 807.0 55.7 34.9 21.3 310.4 16.9 62.4

Figure 3.18 Percentage deviation of each cluster from the overall mean values.

Simple scatterplots which place the cases of each cluster within the axes
defined by selected attributes are also typically employed for visualization of the
clusters' structure. The next set of scatterplots in Figure 3.20 graphically illustrates
the relationship of the clusters in our example with voice and SMS usage. Due to
the large size of the customer population, the scatterplots are based on a random
sample of the assigned records. Moreover, the values of the selected attributes are