You will need to use by group processing to accomplish this. First sort the data by your id and dif_day variable then in a data step you can set the data set with a by id dif_day statement. This creates first.dif_day, first.id, last.dif_day and last.id whose values are 1 or 0 depending on the observation being processed. if the observation is the first of a particular value of the by variable the first. will be 1 otherwise it is 0. So you can use these variables to count which observation in the by group you are processing. Here is a simple example you can cut and paste: data test; do id=1101 to 1103; do dif_day=0 to 20; rtn=int(ranuni(0)*10); output; end; end; run; proc sort data=test; by id dif_day; run; data final; set test; by id; array r(5) (5*0); if first.id then do; counter=0; do i=1 to dim(r); r(i)=0; end; end; counter+1; if counter le 5 then r(counter)=rtn; if counter eq 5 then do; total=1; do i=1 to dim(r); total=total * (1+r(i)); end; total=total-1; output; end; run; ... View more

Here is code to get you started. It will create a data set called missing_vars that has the percent and variable names of the data set test. You will need to macro-tize this so it will run through all the data sets you need to include in your report.   data test; input x1 x2 x3 x4 y $; cards; 1 . 1 1 a 13 2 . 5 b 11 . 1 . c 2 3 4 5 d 3 4 5 6 e . 8 9 . f 5 6 7 8 g 8 9 8 9 h 3 5 8 . i 6 8 7 9 j ; options mprint symbolgen mlogic; data _null_; if 0 then set test; array num(*) _numeric_; array chars(*) _character_; call symput('nums',trim(left(put(dim(num),8.)))); call symput('chars',trim(left(put(dim(chars),8.)))); stop; run; data missing_vars; set test end=end nobs=obsnum; array num(*) _numeric_; array chars(*) $ _character_; array misscountn (&nums.) (&nums.*0); array misscountc (&chars.) (&chars.*0); do i=1 to dim(num); if num(i) = . then misscountn(i) + 1; end; do i=1 to dim(chars); if chars(i) = '' then misscountc(i) + 1; end; if end=1 then do; do i= 1 to dim(num); percentmiss= misscountn(i) / obsnum ; var=(vname(num(i))); output; end; do i= 1 to dim(chars); percentmiss= misscountc(i) / obsnum ; var=(vname(chars(i))); output; end; end; keep percentmiss var; run; ... View more

Important information for using that informat. Thanks. ... View more

I dont think that informat will work here: data test; input date anydtdte.; format date mmddyy10.; cards; 01sep15 01092015 ;;;; run;   test shows two different dates. ... View more

Give this a try.   %macro bias; %let counter=1; %let ok_bias=0; %do %until (&ok_bias=1 or counter gt 100);   data confidentialized; set rollup(Rename=(Members=tmpMembers Spend=tmpSpend)) ; If tmpMembers LE 3 then do ; ... run;   *** BIAS Check ; Proc SQL ; Create table ... quit;   data _null_; set BiasCheck; if memberbias lt .01 and spendbias lt .01 then call symput('ok_bias','1'); run;   %let counter=%eval(&counter + 1);   %end; %put &ok_bias was the ok_bias value; %put &counter iterations occurred.; %mend;   %bias; ... View more