Outputting the first observation

[RandyStan]

Dear All:

My data set is as follows:

Time                  p                  ID

9:00:01              56.57           45

9:00:01              67.97           46

9:00:01              67.97           46

9:00:01               68.00          46

9:00:02              69.00            76

9:00:02              69.10            79

9:00:02              70.01            45

9:00:03              69.04            78

9:00:03              68.10            46

and so on

The output I want is the p observation.  So in this example I need:

Time                  p                  ID

9:00:01              56.57           45

9:00:01              67.97           46

9:00:01               68.00          46

9:00:02              69.00            76

9:00:02              69.10            79

9:00:02              56.57            45

9:00:03              69.04            78

9:00:03              68.10            46

the code I wrote was

data want; set have;

by ID p time:

If first.p then output;

run;

Some thing does not seem to work with the code.

Thanks in advance.

R

[Tom]

Perhaps your values are not actually exactly the same?

[novinosrin]

data have;
input Time  :time8.                p                  ID;
format time time8.;
cards;
9:00:01              56.57           45
9:00:01              67.97           46
9:00:01              67.97           46
9:00:01               68.00          46
9:00:02              69.00            76
9:00:02              69.10            79
9:00:02              70.01            45
9:00:03              69.04            78
9:00:03              68.10            4
;
proc sort data=have out=_have;
by id p time;
run;
data want; set _have;

by ID p time;

If first.p then output;

run;

[Astounding]

What result would you want if the order of p over time is

68
70
68

Should 68 appear twice?

It is possible that this program provides the proper result without sorting

Data want;
Set have;
By p notsorted;
If first.p;
Run;

[mkeintz]

Your sample data appears to be sorted by TIME/P/ID.  if that's true, then your idea of a data step with a BY statement and a"first." condition is correct, but needs tweaking, assuming what you want is to eliminate duplicate time/p/id combinations.

data want;
  set have;
  by time p id;
  if first.id;
run;

Of course this means a given id  or a given P can appears more than once for a given TIME.  I presume that's ok.

[SuryaKiran]

Hello,

Are you trying to remove duplicate records, try the nodupkey option.

proc sort data=have out=want nodupkey;
by id p time;
run;