Segmentation Applications in Banking - Data Mining Techniques in CRM: Inside Customer Segmentation - page 242

Figure 6.5 Bubble plot illustrating the percentage of purchases at supermarkets,
clothing stores, and telecommunication operators per cluster.
Y -axis co-ordinate the spending on telecommunication services, and the dot size
the spending at supermarkets. The reference lines correspond to the overall means
and hence represent the ''average'' credit card holder.
The bubble plot outlines the relation of cluster 6 to purchases of clothes/
accessories and the association of cluster 2 with telecommunication services. Their
percentage of purchases at relevant merchants is by far the largest of all clusters.
The larger size of the dot corresponding to cluster 3 designates the high spending
of this cluster at supermarkets. The remaining clusters are not clearly differentiated
by the three usage fields summarized in this plot.
An analogous bubble chart is presented in Figure 6.6, summarizing the derived
clusters in terms of relative spending on entertainment, traveling, and home/
garden products.
Cluster 5 has the highest percentage of spending at home stores. Cluster 7 is
associated with entertainment and cluster 4 with travel expenses. In both bubble
plots, cluster 1 is close to the reference lines and shows some proximity to the
average profile.
The next step, after the exploration of the relative spending by merchant
categories, was the examination of possible associations of the clusters with other