I was trying to come up with this but didn't know about the intnx function (thanks!) so had to go the long way round through multiple queries. This is neat. ... View more

For record 12, you show a count of 3 in the desired results.    Should that be 4?  I think records 9, 10, 11, and 12 all fall within the 3 day time frame, but I may be wrong. Record 8 is too far back and Record 13 is too far forward.  It may be an < vs <= thing.  ... View more

I don't open strange files from people I don't know.    Just post text or code.  ... View more

Or if you want you can kick it to another statement entirely just for fun.   Data vitals ; input ID $3. Age Pulse SBP DBP; Datalines ; 001 23 68 120 80 002 55 72 188 96 003 78 82 200 100 004 18 58 110 70 005 43 52 120 82 005 43 52 120 82 006 37 74 150 98 007 . 82 140 100 ; data vitals; set vitals; label SBP = "Systolic Blood Pressure" ; label DBP = "Diastolic Blood Pressure" ; run; ... View more

" The planname used to be numeric, but it is now alphanumeric, which is causing the error. "   What error is that?   If you had a field you were using as a unique identifier it should have been character to begin with- you aren't going to add or subtract it, it isn't numeric even if it was all numbers.    Maybe you could explain more about what the problem is? ... View more

I think making that   '/:?'   gets OP to the desired result (dropping questions marks).   Although I do think 16/nq becoming 16 and nq while n/q becomes " " and nq  is a little strange.    ... View more

I don't think that will split 1000mg into 1000 and mg. ... View more

Should the last record be    9     shan     male     4     8.1          1  1   1.6789      2     3   or does the : instead of the / mean disregard what is after?     If you are looking to get rid of ?'s what should this records look like?   5     gita       female  2     2.1       99 95    1.23456   ?        prp   Should it be  5     gita       female  2     2.1       99 95    1.23456             prp   ... View more

EDIT:  I JUST SAW THIS HAS BEEN DISCUSSED AND SOLVED IN ANOTHER THREAD.  DON'T DOUBLE POST UP LIKE THAT.  I'LL LEAVE THIS ANSWER IN CASE IT APPLIES TO SOMEONE ELSE'S PROBLEM.       You could do it less elegantly with a subquery.   You don't post any sample data and not real good guidance but given data like     Data aa; input id one mydate:mmddyy10. something:$6.; format mydate yymmdd10.; datalines; 111 100 6/1/2014 yergle 112 200 7/2/2014 omnipt 113 500 8/4/2014 weezle ; run; Data bb; input id one mydate:mmddyy10. var1:$3. var2:$3. var3:$3.; format mydate yymmdd10.; datalines; 111 100 7/1/2014 AAA bbb ccc 111 200 5/30/2014 bbb ccc ddd 111 500 5/15/2014 ddd eee fff 112 100 7/3/2014 AAA xxx sss 112 200 7/4/2014 AAA zzz qqq 112 500 7/5/2014 ABC vvv bbb 113 100 6/1/2014 AAA nnn mmm 113 200 6/2/2014 AAA eee ttt 113 500 6/4/2014 ABC fff ggg ; run; if I understand correctly you would want data from BB records with id of 111 and var values of bbb ccc ddd and 113 ABC fff ggg   A subquery could find the closest record and then you can join back.   proc sql; select bb.*, aa.something from ( select bb.id, max(bb.mydate) as myccdate from bb inner join aa on bb.id = aa.id where bb.mydate <= aa.mydate group by bb.id ) as cc left join bb on bb.id = cc.id and bb.mydate = cc.myccdate left join aa on bb.id = aa.id; quit;   This produces     The SAS System id one mydate var1 var2 var3 something 111 200 2014-05-30 bbb ccc ddd yergle 113 500 2014-06-04 ABC fff ggg weezle       Maybe this is completely off base.  Maybe it will be useful.    ... View more

Still don't know what you want or what you are doing.    Just guessing a this point that instead of   from temp.database,maps.Storelist where Storelist.ID=Database.ID   you may want   from temp.database maps.Storelist inner join Storelist.ID=Database.ID where database.store in (002,005,008) In the IN clause you may need 2, '2', '002', or something.  Don't know how your data is stored.  ... View more