Create a flow direction in proc sql

[mkeintz]

I wrote my program to accommodate any sequence of links within an ID group.  Although the example had each link descend from the prior record, the OP didn't state that as a condition.  But if that condition can be guaranteed to always be the case then the program is much simpler.   Just copy the sequence of FROM values and append the final TO value for each ID.  Then transpose:

data have;
  input id  from :$1. to :$1.;
datalines;
1 a b 
1 b c 
1 c d 
2 e f 
2 f g 
run;

data vneed / view=vneed;
  set have;
  by id;
  seq+1;
  if first.id then seq=1;
  if first.id then nodename='Source';
  else nodename=cats('Node_',seq);
  output;
  if last.id ;
  from=to;
  nodename='Sink';
  output;
run;

proc transpose data=vneed out=want (drop=_name_);
  by id;
  var from;
  id nodename;
run;

[Ksharp]

No. It can be settled randomly.
Or have a dead loop like:


ID	from	to
1	a	b
1	b	c
1	c	d
1      d      a