Data step reatin first non zero value by group

[sivakoya]

ID	value
1	0
1	0
1	100
1	200
2	0
2	0
2	200
2	0
2	0
3	0
3	100
3	50
3	0
3	200

Hi ,

 My data looks like this. Variable 'value' is an amount field and its a LTD field, once it gets populated it should never decrease or become zero. Dollar amount in the 'value' field could only grow. So the requirement is to retain greater value in each by group (ID).

desired output:

ID	value	desired
1	0	0
1	0	0
1	100	100
1	200	200
2	0	0
2	0	0
2	200	200
2	0	200
2	0	200
3	0	0
3	100	100
3	50	100
3	0	100
3	200	200

My dataset has 153 million records, is there an easy way to do this using data step functions?

[ballardw]

data have;
 input ID  value ;
datalines;
1 0 
1 0 
1 100 
1 200 
2 0 
2 0 
2 200 
2 0 
2 0 
3 0 
3 100 
3 50 
3 0 
3 200 
;
run;

data want;
  set have;
  by id notsorted;
  retain  desired;
  if first.id then desired=0;
  if value gt desired  then desired=value;
run;

I used the NOTSORTED option for your ID variable as it is not clear whether the data is actually sorted by ID or that ID may occur in different sequences of records.

[Ksharp]

data have;
 input ID  value ;
datalines;
1 0 
1 0 
1 100 
1 200 
2 0 
2 0 
2 200 
2 0 
2 0 
3 0 
3 100 
3 50 
3 0 
3 200 
;
run;
data want;
 set have;
 by id;
 retain want . ;
 if first.id then call missing(want);
 want=max(want,value);
run;