Identify independet "network clusters"

[user24feb]

Hello!

I would like to split network-data containing "from" and "to" items into independent "clusters". For example:

Data Have;
  Input From To;
  Datalines;
200 401
200 460
460 402
460 403
461 400
299 500
299 501
500 508
500 509
;
Run;

Data Want;
  Input Cluster_Nr Members $;
  Datalines;
1 200_460_461_403_402_401_400
2 299_500_501_508_509
;
Run;

This is, there is no path leading from 299 to 400 or from 200 to 509, etc.

My 2 questions are:

1) is there a standard way, this is using for example a SAS/OR procedure, to do this?

2) if there is no standard way to do this, how could it be done?

Thx&kind regards

[PGStats · Accepted solution]

Look for the statement CONCOMP in PROC OPTNET of SAS/OR. It finds connected components.

SAS/OR(R) 12.3 User's Guide: Network Optimization Algorithms

PG