To bring this all together, consider the following example, from the Tech Support domain. Figure 4.13 reproduces Figure 4.3 here for convenience. In fact, Figure 4.3 did in fact use K-Means clustering to determine the centroids.
Centroids are marked with diamonds. Also note that a simple Euclidean distance metric is used, and different standard deviations for the different axes are not taken into account (hence the results look a little unusual). Also note that a priori we specified three clusters. The dotted lines represent the region boundaries; these are easy to compute as they are the loci of points that are equidistant from the two nearest centroids.
