1 / g for that species to be included in the cluster. This formula is such that Smin decreases as the size of the cluster increases, thus allowing the largest clusters to be more heterogeneous.

The clustering steps are shown in Table 8.10 for the cluster nuclei (1, 2) and (5, 6) found during the first clustering step. In each case, the clustering procedure stops when no further significant similarities are found. The two clusters could have had species in common, although it is not the case here. At the end of this first clustering cycle, only one of the clusters is retained, i.e. the one containing more species; in cases of equality (as is the case here), the cluster that has the highest internal probability is selected. The internal probability of a cluster is the product of the complements of the similarities (1 - S) between all members of the cluster. It is shown in Table 8.10 for the two clusters formed. The four species (5, 6, 7, 8) forming the cluster with the

Table 8.12 Similarity matrix for the two clusters of 4 species. In both cases, the highest similarity value Smax (in italics) does not reach the minimum significant value, so that the similarity matrices are considered homogeneous and the clustering procedure stops.


Was this article helpful?

0 0

Post a comment