Based on this first result, how would you describe the data set? Do you see a
single simple gradient, or else distinguishable groups of sites? Can you identify
some chaining of the sites? What about sites 1, 5 and 9?
4.3.2 Complete Linkage Agglomerative Clustering
Contrary to single linkage clustering, complete linkage clustering (also called
furthest neighbour sorting) allows an object (or a group) to agglomerate with another
group only at the dissimilarity corresponding to that of the most distant pair of
objects; thus, a fortiori, all members of the two groups are linked (Fig. 4.2):
1
12
2
3
7
11
13
14
9
15
16
10
4
6
5
19
17
18
29
30
20
21
22
26
27
28
25
23
24
spe.ch
hclust (*, "complete")
Height
1.4
1.2
1.0
0.8
0.6
0.4
0.2
Chord - Complete linkage
Fig. 4.2 Complete linkage agglomerative clustering of a matrix of chord distance among sites
(species data)
64
4 Cluster Analysis