|
I remember I've used n*S where S is the Sokal's binary distance.
The advantage is that you can enter your binary data directly in the
PROC CLUSTER because n*S is equal to Euclidian distance.
If my memory is working well , PROC CLUSTER can handle
distance data instead of raw data. Then you can use MACRO
DISTANCE to compute distances from several ways. If you
are using SAS v9.? then there is a PROC DISTANCE to do this.
HTH,
Rogerio Porto.
|
