I always couldn't get it, why SAS has to pair "parmcards4" with an obscure device name such as "FT15F001"? It would be nice if SAS can remove this kind of restriction, so programmers don't have to remember or  look it up in docs of version 8 and just come up some generic names they like. ... View more

You already have your answer, both answers from @ChrisNZ and @Ksharp should work. I am here just to comment that your original example code will not work for your purpose. The reason being SAS using FIXED length variable, you will need byte abundancy or some walk-around (rename etc.) to replace with a longer chars, notice @Ksharp's example with some blanks padded.   ... View more

This can be solved using Hash, however, make sure you don't have infinite loops, if you do, set a loop limit (say the total number of your obs), when reached, exit the loop regardless.   data have; input ID1 ID2 ; cards; 11 22 22 35 35 9 41 10 52 87 9 65 ; run; data want; if _n_=1 then do; dcl hash h(dataset:'work.have'); h.definekey('id1'); h.definedata(all:'y'); h.definedone(); call missing(id1, id2); end; set have(rename=(id1=_id1 id2=_id2)); length id3 $20.; rc=0; id1=_id1; do while (rc=0); rc=h.find(); id1=id2; end; id3=catx('_', 'ID', PUT(id1,BEST.)); KEEP ID3 _id:; run; ... View more

On top of @Tom's suggestion, If your variables of interest share some traits, such as they are at least 2 with the same prefix (shown in the example code ), or all variables except 'ID'/'Group' etc.(not shown),  you could use some variants of the following to save some typing:   %macro atleast(list,min=2); %local i base; %do i=1 %to %sysfunc(countw(&list)); %let base=%scan(&list,&i); if n(of &base.:) >= &min then &base._avg=mean(of &base.:); %end; %mend atleast; proc sql; select DISTINCT PRXCHANGE('s/\d*$//',1,strip(name)) into :list separated by ' ' from dictionary.columns where libname='WORK' AND MEMNAME='HAVE' group by PRXCHANGE('s/\d*$//',1,strip(name)) having count(*) >=2 ; QUIT; data want; set have ; %atleast(&list.) run; ... View more

I would assume you are using local SAS 9.1+  (not server version), and you have access to your local SAS folder. 1. Locate sasv9.cfg you can just simply search it using OS or right click SAS icon check on properties, you may see something like: "C:\Program Files\SASHome\SASFoundation\9.4\sas.exe" -CONFIG "C:\Program Files\SASHome\SASFoundation\9.4\nls\en\sasv9.cfg" , that tells you where to locate the config file sasv9.cfg 2. Using text editor (e.g. notebook) open it and look for :-WORK “c:\whatever\”, then change it to where you want to route your temp file. Save the config file. restart SAS. ... View more

data dat2; set dat1; if time=. then output; else if time ge 0 then do count=1 to round(time, 1); output; end; run; ... View more

You didn't mention if your Eminer web based or client based. If web based, I would close all the browser window and clean up Java cache before calling my SAS admin.   Java cache cleaning     ... View more

data new; input CustID TranAmt:dollar6. TranDate:mmddyy10.; format TranDate mmddyy10.; datalines; 111 $100 6/1/2014 111 $200 6/2/2014 111 $500 6/4/2014 111 $500 6/10/2014 111 $100 6/11/2014 222 $200 6/3/2014 222 $500 6/4/2014 222 $600 6/4/2014 222 $600 6/11/2014 222 $600 6/19/2014 222 $600 6/28/2014 222 $600 6/29/2014 ; run; /*hwo about only loading one custid at a time, following tweak on Ksharp's code should circumvent or at least alleviate ram stress*/ data want_hash; if _n_=1 then do; declare hash h(multidata:'y', ordered:'y'); h.definekey('TranDate'); h.definedata('TranDate','_TranAmt'); h.definedone(); call missing (_TranAmt); end; do until (last.custid); set new; by CustID TranDate; if first.Custid then rc=h.clear(); rc=h.add(key:trandate,data:trandate, data:tranamt); end; do until (last.custid); set new; by CustID TranDate; sum_8day=0; count_8day=0; do i=TranDate-7 to TranDate; rc=h.find(key:i); do while(rc=0); sum_8day+_TranAmt; count_8day+1; rc=h.find_next(key:i); end; end; output; end; drop i rc _TranAmt; run; /*this will slow, but a lot easier to code and read*/ proc sql; create table want_sql as select *, (select sum(tranamt) from new a where a.custid=b.custid and b.trandate-a.trandate between 0 and 7) as sum_8days, (select count(*) from new a where a.custid=b.custid and b.trandate-a.trandate between 0 and 7) as count_8days from new b ; quit; ... View more