Yes, function indexc() searchs for one single character in the string, where index() searchs for the whole text.  You can compress your code somewhat, if <> is looking for a binary 1 or 0 result, so index() or findw() (which i show below) will return 0 if not present, so 0 result, and anything is positive and considered true (also, please dont code all in capitals, it makes reading it very hard): data test; input name $50.; datalines; DREFT ESTATES LTD POTTERSLANE LTD ; run; data for_sure unsure; set test; if findw(name,"GOVERNMENT") or findw(name,"COLLEGE") then output for_sure; else output unsure; run;   ... View more

Data test; input Id a b c; infile datalines; datalines; 123 20 30 50 124 1 5 8 ; run; data want; set test; array x{*} a b c; do j=1 to dim(x); total=vname(x{j}); do i=1 to x{j}; output; end; end; drop i j a b c; run; ... View more

Thanks Miguel, this is really helpful. I do have Naeem Siddiqui book and familiar with vintage analysis. I understand your ideas ... View more

So if you have 1000 defined variables you should store that information in a structured way, rather than as macro variables.  You could create a table that looks like the output of PROC CONTENTS with variables for NAME, TYPE, LENGTH, FORMAT and LABEL.  Or you could just create a representative table and use PROC CONTENTS on that table to generate the data set with the metadata. Once you know the types that the variables should have then you can skip using PROC IMPORT and instead generate the code to import the file yourself.   If the issue is that each file might have only some of the 1000 possible variables and/or that the order of the variables changes then just read the first line of the file to find the names and merge it with your metadata and generate the code to read the rest of the file.  This could in fact be very simple.  Let's assume that your raw input file is called "mydata.csv" and that you have a sample file with all of the possible variables named SASDATA.TEMPLATE.  To use that to generate the code to read "mydata.csv" into SASDATA.TEST would be very simple since you just need to read the first line and convert it into INPUT and KEEP statements. filename code temp; data _null_;   infile "myfile.csv" lrecl=300000 obs=1;   input ;   _infile_ = translate(_infile_,' ',',');   file code ;   put 'input ' _infile_ ';';   put 'keep ' _infile_ ';' ; run; data sasdata.test ;   if 0 then set sasdata.template;   infile "myfile.csv" lrecl=300000 dsd truncover firstobs=2 ;   %inc code / source2 ; run; ... View more

Don't be shame of it ,Jaap. SAS is so huge, you can't remember every details of it , and further more , SAS add lots of function proc option  in every SAS version release .That is why we need keep learning -> remember -> forgotten ->re-remember -> forgotten again  ......  It is revolution . Xia Keshan ... View more

Hi, As your data is currently normalized, why not use SQL to get the information: data test; infile datalines; input ID TIME monyy5. Point; datalines; 1 Jan10    1 1 Feb10    2 1 Mar10    2.1 1 Apr10    2 1 May10    2 1 Jun10    2.1 1 Jul10    2 1 Aug10    2.1 2 Jan10    1 2 Feb10    1 2 Mar10    2 2 Apr10    2.1 2 May10    2.1 3 Jan10    1 3 Feb10    2 3 Mar10    2.1 3 Apr10    3 ; run; proc sql;   create table WANT as   select  BASE.ID,           BASE.TIME,           BASE.POINT   from    WORK.TEST BASE   where   BASE.POINT - int(BASE.POINT) > 0     and   EXISTS(select distinct                         THIS.ID                  from   WORK.TEST THIS                  where  THIS.ID=BASE.ID                   and   THIS.TIME > BASE.TIME); quit; The above only gives you the actual y.x result, but its straight forward to find the next or previous row = min(time) > base.time. ... View more

data have; input var1 $ var2 var3 :date9.; format var3 date9.; cards; A    .    . A    .    . A    .    . A    1    16APR2010 A    .    . B    .    . B    1    17MAY2010 B    .    . B    .    . ; run; /*Distinct number of values in var1*/ proc sql; select count(distinct var1) into: num from have; quit; %put &num; data want; do i=1 by 1 until(last.var1);     set have;     by var1;     retain count;     if var2=1 then count=i; end; do i=1 by 1 until (last.var1);     set have;     by var1;     if i<=count then output;     end; format var3 date9.; drop i count; run; ... View more

You are getting an error that your personal proifles (that is sasuser) is not accessible. When you are using Eguide to a server to a SAS workspace server (or other) this is quite common behavior. Whe you are using SAS base on a PC it looks a definiton has been corrupted. It should point to a "My Documents" location.  Conforming the Microsoft requirements and settings. The C:\work  advice is one of possible ancient times as C: should not be used in that way, closed down. Also for the work there are microsoft guidelines mostly pointing to a local-settings submap (personal) as Windows is Multi-user Multi-process.. Please give more information on your situation.      ... View more

"SAS Data Integration Studio" is the application, or software if you like. ... View more

This Discussion is worth reading Regarding macro Global&local variables  Thanks all.. ... View more

I think your question is how can you read the variable length string in the middle of the line that might had embedded blanks and does NOT have two spaces after it to indicate the end of the variable length string. In general you cannot.  So you will have to use some other method to determining where the number at the end of the line starts. DATA test;   INPUT  x y a & $14. ;   j=input(scan(a,-1,' '),2.);   a=substr(a,1,length(a) - length(scan(a,-1,' ')));   put / _infile_ / (_all_) (=); DATALINES;                1 2 HELLO WORLD 3               4 5 S BEHNOI 6   ; 1 2 HELLO WORLD 3 x=1 y=2 a=HELLO WORLD j=3 4 5 S BEHNOI 6 x=4 y=5 a=S BEHNOI j=6 ... View more