data try_1; input ID     c_ID     d_status     age     fin_grp; datalines; 1    1    1    6    1 1    2    1    12    1 1    3    1    15    1 2    1    2    10    0 2    2    2    14    0 2    3    2    16    0 4    1    1    6    1 4    2    2    13    1 4    3    2    15    1 5    1    1    9    0 5    2    2    12    0 5    3    2    11    0 ; data want (drop=age1 c_id1); age1=.; do until(last.id); set try_1; by id; where (fin_grp=1 and d_status=1) or (fin_grp=0 and d_status ne 1); age1= min(age1,age); if age<=age1 then c_id1=c_id; else if age>age1 then c_id=c_id1; end; age=age1; run; ... View more

data have;      infile cards truncover;      input period    value;      cards; 1 10 2 5 3 4 12 ; data want; set have; retain value1; if value ne . then do; value1=value; return=(value-lag(value1))/lag(value1); end; drop value1; run; ... View more

Sorry, Your question is not quite clear or my understanding is poor. I can only agree with Reeza on looking at your data. Can you explain to Reeza's point "It looks like you want to keep records with multiple occurrences rather than remove duplicates?" . ... View more

PG is basically using that as a index or in other words count variable to retireve each word in the string one by one as the scan function does scan(string,count) where count can be number, variable or a derived expression. so in your case he is using index variable as count to extract starting from my, then cat and so on. You could also use something like: do until(word=' ');   count+1;   word = scan(string, count);    if not missing(word) then output;   end; Just different ways to write the code. Regards, Naveen ... View more

data _null_; date1='01JAN14'd; date2='01OCT14'd; x = intck("month",date1,date2); y=month(date2)-month(date1); put x= y= ; run; ... View more

If using a datastep: data want; set have; Name=vname(abc); Age=abc; drop abc; run; ... View more

I am novice, unlike the greats PG, Xia and Haikuo, hence the long and ugly way, for learning purpse it helps: data have;/*Raw data copied from Haikuo*/      input subjid : $20. number $ cat & $40. visitnum;      cards; 100010001           T02         TARGET LESIONS              0 100010001           T01         TARGET LESIONS              0 100010001           T03         TARGET LESIONS              0 100010001           T02         TARGET LESIONS              1 100010001           T03         TARGET LESIONS              1 100010001           T01         TARGET LESIONS              1 100010001           T02         TARGET LESIONS              2 100010001           T01         TARGET LESIONS              2 100010001           T03         TARGET LESIONS              2 100010002           T02         TARGET LESIONS              0 100010002           T01         TARGET LESIONS              0 100010002           T02 TARGET LESIONS              1 100010002           T01         TARGET LESIONS              1 100010002           T02         TARGET LESIONS              2 100010002           T01         TARGET LESIONS              2 100030002           T03         TARGET LESIONS              0 100030002           T04         TARGET LESIONS              0 100030002           T02         TARGET LESIONS              0 100030002           T01         TARGET LESIONS              0 100030002           T01         TARGET LESIONS              1 100030002           T04         TARGET LESIONS              1 100030002           T03         TARGET LESIONS              1 100030002           T02         TARGET LESIONS              1 160020001           T02         TARGET LESIONS              0 160020001           T01         TARGET LESIONS              0 160020001           T02         TARGET LESIONS              1 ; run; data want; set have; by subjid visitnum notsorted; array n(30) _temporary_; *array subscript is arbitrary; if first.subjid then call missing(of n{*}); if first.visitnum then count=0; count+1; if last.visitnum then n(_n_)=count; if last.subjid then do; if max(of n ) ne count then discrepancy="yes"; else discrepancy="no"; end; if discrepancy="yes" then output; drop count; run; proc sort data= have out=have1; by subjid; run; data final; merge  want have1; by subjid; if discrepancy="yes" then output; drop discrepancy; run; Regards, Naveen Srinivasan L&T Infotech ... View more

Hi, &path will hold the value of the path. So if you are specifying the full path name, don't use macro &path libname libref "your path"; *this is the syntax; try this:  libname custfm 'C:\Users\cflach\Documents\custfm.xls' ; * assuming this is where your xls file is located; ... View more

Hi, I believe the reason why C.Torres suggested the use of proc datasets is because you can avoid datastep processing like in your example writing to a new dataset. You are pretty much compiling, executing and doing the whole process that is something can be avoided. You can assign a format even with an attribute statement, nevertheless proc datasets is a powerful procedure for managing SAS datasets. HTH, Naveen ... View more

I don't think you need to specify pcfiles engine and path= coz &path will just resolve to the defined path of macro by the macro processor by substituting it's value from the global symbol table and places it back into the input stack. libname custfm "&path\custfm.xls"; *this should work getting your libname assigned. ... View more

Haikuo's codes are classy and expert level, but if you are learner like me, you can join my club with simple code below, let's learn together from haikuo data have;/*copied haikuo's raw data*/      input (CLIENT   PRODUCT) (:$8.) NUMBER;      cards; Jane a    1 Jane b    2 Jane b    2 Angie a    1 Angie a    1 Angie b    2 Angie c    3 Lisa c    3 Lisa b    2 data want; set have; by client product notsorted; retain n; if  first.client and first.product then n=product; else if  not first.client and product="b" and n="a" then number=5; drop n; run; ; ... View more