@novinosrin: No, the traffic there has been diminishing for a number of years now. However, I still look there every day as it doesn't require a lot of time and I still learn things that I otherwise would have missed.   Art, CEO, AnalystFinder.com   ... View more

This question was cross posted on the SAS-L bulletin board. There, @data_null__ showed a method that is way more efficient, namely one that doesn't require reading and re-writing all of the records you want to keep: data have; point=nobs; modify have nobs=nobs point=point; remove; stop; run; Art, CEO, AnalystFinder.com   ... View more

@A_SAS_Man: I would simply rename them within the data parameter and use the same code. e.g.: data current_state; input id $ att $ type $ def $ part $ section $ Dollars; datalines; 1 d234 d463 d213 d678 d435 50 2 d213 d690 d360 d145 d269 25 3 d409 d231 d463 d690 d609 10 ; filename ut url 'http://tiny.cc/untranspose_macro'; %include ut ; %untranspose(data=current_state (rename=(att=att1 type=att2 def=att3 part=att4 section=att5)), out=long, by=id, var=att, id=na, copy=Dollars) filename tr url 'http://tiny.cc/transpose_macro'; %include tr ; %transpose(data=long, out=want, by=id, var=Dollars, id=att, sort=yes, use_varname=no) Art, CEO, AnalystFinder.com   ... View more

Should be away to do it with a regular expression but, again, let's wait to see what @atul_desh really wants.   Art, CEO, AnalystFinder.com   ... View more

@novinosrin: If your assumption is correct your code only needs one escape statement. e.g.: data ds2; set ds1; array fc[*] fc_1-fc_10; array dt[*] dt_1-dt_10; do i=dim(fc) by -1 until(fc(i-1) eq 0 and fc(i)>0); if fc(i-1) eq 0 and fc(i)>0 then do n=i-1 by -1 while (fc(n) eq 0); stp_dt=dt(n); if n eq 1 then leave; end; end; drop i n; run; Art, CEO, AnalystFinder.com   ... View more

@novinosrin: You're assuming that @atul_desh didn't just make an error regarding what was desired for the 4th test case. That may or may not be the case.   If it is, your code as is won't correctly handle a case like: 0 0 0 4 3 10 1 9 4 7 a b c d e f g h i j Art, CEO, AnalystFinder.com     ... View more

If your fourth value should really be c rather than b, then the following would work: data ds1; input fc_1-fc_10 (dt_1-dt_10) (:$2.); datalines; 7 0 8 6 5 0 7 0 0 0 a b c d e f g h i j 7 0 3 10 3 2 10 8 4 7 a b c d e f g h i j 3 0 0 0 2 7 4 4 0 9 a b c d e f g h i j 3 0 0 4 3 10 1 9 4 7 a b c d e f g h i j ; data ds2; set ds1; array fc[*] fc_1-fc_10; array dt[*] dt_1-dt_10; do i=dim(fc) by -1 until(fc(i-1) eq 0 and fc(i)>0); if fc(i-1) eq 0 and fc(i)>0 then stp_dt=dt[i-1]; end; run; Art, CEO, AnalystFinder.com   ... View more

@sovrappensiero: The SQL part is simply building code for the data step that follows it. proc sql noprint; select catt(var,'=_',var), catt(var,'=input(_',var,',8.);') into :rens separated by ' ', :vars separated by ' ' from owf ; quit; The first catt in the select section is building a space separated list to recode. I.e., given var1 and var2, e.g., it is creating: var1=_var1 var2=_var2, and putting the result in a macro variable called &rens.   The second catt is building a set of statements that will take advantage of those recodes. i.e., given var1 and var2, e.g., it is creating var1=input(_var1,8.); var2=input(_var2,8.); and putting the result in a macro variable called called &vars.   The data step that follows simply uses the two macro variables as code to recode the variables and, for the latter, statements that will accomplish the actual conversion from character to numeric.   Art, CEO, AnalystFinder.com   ... View more

Unfortunately you can only select one correct solution. However, there aren't any community rules regarding which one you select .. i.e., the choice is yours. You can always 'like' as many responses as you care to mark that way, but only one can be marked as being 'correct'.   Art, CEO, AnalystFinder.com   ... View more