Figure 1

Illustration of the definition of tightness. The data consist of 280 points in one dimension, drawn from a normal mixture with the components N(0.5,0.4²) (blue), N(11,1²) (green) and N(5,2²); (black). A) A histogram of the input data. B) A hierarchical tree of the input data, grown using the absolute difference of the data values as the dissimilarity measure, and single linkage. Thus, the node heights shown in (B) are equal to the corresponding gaps in the data, as indicated in (A). Nodes n₁ and n₂ are approximately equally tight.