|
|
Hi all.
The title explains my question.
Seems like a very simple question, but I can't figure it out. I have
carried out a very simple clustering problem using hierarchical clusterng
for 4 objects - a, b, c and d. Each object has 2 measures as displayed:
a: 3 4
b: 7 6
c: 5 1
d: 1 9
I used nearest neighbour and Manhattan distance.
The clutering results are as follow
Stage Cluster_Combined Coefficients
1 a&c 5
2 a&b 6
3 a&d 7
I understand the interpretaton of the above table, but the plotted rescaled
dendrogram shows a&c initially being linked at what looks like 1 on the
scale, {a&c} & b being clustered around 12 and the final cluster occuring
at 25. How does SPSS scale from coefficients (which are simply the
Manhattan distances between clusters) to the values of (approximately) 1,
12 and 25?
Help appreciated.
Am I missing something very simple or misinterpreting this?
Regards,
Barry
|
Locked
