Dividing summary statistic by 1 million in Proc Summary Step

[Terho]

Hello,

I am using a proc summary step to summarize my data.

However, I want to divide the summary statistics in the "VAR" statement below by 1,000,000.

PROC SUMMARY DATA=ExecSummary nway missing;
CLASS
JOBID
PERIOD_ID
SCENARIO;
VAR
RQ01
RQ02
NCO_9Q
NCO_13Q
;
OUTPUT out=ExecOutput (drop=_FREQ_ _TYPE_) sum=;
RUN;

Is there anyway to divide the summary statistics by 1 million directly in the proc summary step?

Thanks

[Reeza]

Can you apply a format to the results instead to scale the results?

---

@Terho wrote:

Hello,

I am using a proc summary step to summarize my data.

However, I want to divide the summary statistics in the "VAR" statement below by 1,000,000.

PROC SUMMARY DATA=ExecSummary nway missing;
CLASS
JOBID
PERIOD_ID
SCENARIO;
VAR
RQ01
RQ02
NCO_9Q
NCO_13Q
;
OUTPUT out=ExecOutput (drop=_FREQ_ _TYPE_) sum=;
RUN;

Is there anyway to divide the summary statistics by 1 million directly in the proc summary step?

Thanks

---

[mkeintz · Accepted solution]

You could do make a weight variable:

data need / view=need;
  set execsummary;
  _wgt=1/1000000;
run;

PROC SUMMARY DATA=need nway missing;
  CLASS  JOBID  PERIOD_ID  SCENARIO;
  VAR  RQ01  RQ02  NCO_9Q  NCO_13Q ;
  weight _wgt; 
  OUTPUT out=ExecOutput (drop=_FREQ_ _TYPE_) sum=;
RUN;

There is not much efficiency loss here, since NEED is a data set view, not a data set file.  NEED will be activated only when called by the proc summary, and will stream its data to the proc, not to disk storage.  So virtually no input/output cost. 

[Terho]

Thank you! This worked wonderfully!