Remove the macro call that you make immediately after the definition. You also have a comma after ddate. (the second one).   see what happens after removing that ... View more

First thing I see if that you execute the macro directly after the definition unnecessarily.  Please include a log of the error. ... View more

Break your problem into steps.   First you need to determine the length of the longest observation in TEST variable.  Once you have that, you can then scan across each observation (by 200) checking if the character sitting in the 200n+1 position is a space (if not, find the closest space in an earlier position). I’d recommend looking into array processing for this step.  ... View more

I'd take this approach:   data want; format mydate date9.; do i='14AUG2019'd to '16SEP2019'd; mydate=i; if day(mydate) in (2,5,10,15,20,25) then output; end; drop i; run; proc sql noprint; select "'"||put(mydate,date9.)||"'d" into :maclist1 separated by "+" from want ;quit; %put &maclist1; proc sql noprint; select put(mydate,yymmn6.) into :maclist2 separated by "+" from want ;quit; %put &maclist2; ... View more

Hi there. Thank you for your detailed post!   Try something like this out:   data have; input client dx_cd1 $ dx_cd2 $ dx_cd3 $ dx_cd4 $ dx_cd5 $; datalines; 1 291 2911 2915 2911 2915 2 292 2923 2924 2924 2924 3 F101 F102 F110 303 291 4 Z345 F181 F191 Z235 Z786 5 Z345 Z235 Z126 Z786 Z235 6 305 304 303 303 2915 292 7 F191 Z786 F161 F102 F101 8 313.1 Z235 Z126 Z786 Z235 ; run; data want; set have; flag=0; array dxarr dx_cd:; do over dxarr; if substr(dxarr,1,1)='F' or compress(dxarr,'.','sd')='' then do; flag=1; leave; end; end; run;   Basically the idea is that you search over your dx_cd vars and if any of them are in your list, you move onto the next observation.   The way compress is used here, I remove digits ('d'), spaces ('s') as well as '.' (as @ballardw pointed out, many ICD10 codes contain '.' -- notice that here client 8 receives flag=1). Since these are the only "flaggable" characters, we want to flag only those that come up empty (null) after removal. ... View more

It looks like the index allows you to discern between different holidays that have fallen on the same date. In @novinosrin's example, Veteran's day is observed in three different holiday schemes: Julian/calendar, US Gov., Postal (USPS). The index is somewhat unpredictable for holidays that depend on whether or not the true calendar day falls on a weekend (US Government holidays will always fall on a weekday -- fantastic if you work in the Public sector :)). See here:   data test; length holiday $32; format date weekdatx17.; do date='01jan2017'd to '31dec2019'd; cnt=holidaycount(date); do index=1 to cnt; holiday=holidayname(date,index); if holiday in ('VETERANS' 'VETERANSUSG') then output; end; end; run;   If you want only specific holiday dates for a given time range, you might scan across the years of interest and then pull out the holidays you want by name.   -unison ... View more

Perhaps model the log() of your desired model and then it becomes some sort of linear-log model? Just a thought.   -unison ... View more

Here's an approach using a data _null_ step:   data _null_; set sashelp.class(obs=1) indsname=dsname; array num_vars(3) Age Height Weight; do i=2 to dim(num_vars); call execute(cat( 'proc corr data=',dsname,' out=','output_',vname(num_vars(1)),'_',vname(num_vars(i)),'; ', 'var ',vname(num_vars(1)),' ',vname(num_vars(i)),';', 'run;' ) ); end; run;   -unison ... View more