Did you ever find a cheat sheet of several most used dates and such? The list of all formats is a bit cumbersome. ... View more

This may be useful to someone. I was looking to control SAS processing, based on the value of _ERROR_. If the value of error can be converted into a macro variable, then this becomes possible.  The code is as follows:   proc delete data = have; run;   data have; set sashelp.class; call symputx(' error ', _error_); run;   %put &error;   proc delete data = have; run;   data have; set sashelp.class; v = 10 + 'GGGGGG'; call symputx(' error ', _error_); run;   %put &error;   proc delete data = have; run;   data have; set sashelp.class; call symputx(' error ', _error_); run;   %put &error; ... View more

That Alt+LftMouseClk was very sweet, many thx!! ... View more

Hi Cynthia,   Your program:   proc tabulate data=sashelp.class f=comma6.0; title '2) PROC TABULATE equivalent of MEANS output'; var _numeric_; table _numeric_, nmiss n / box='Variable'; run;     count non-missing values for numerical variable, I also want count for character variable, could help me modify this code to achive my goal?   Thanks a lot!   Best wishes, ... View more

Magic! Thank you so much! ... View more

NULL and MISSING, while very similar, have some semantic differences and use of WHERE statements/clauses can produce different results depending upon the engine (e.g., BASE, ACCESS).  The following test case demonstrates this: /*    Test showing that a WHERE applied to a DBMS table via    an ACCESS engine will produce results different from those    produced when the WHERE is applied to a SAS data set.    The difference is caused by the treatment of NULL/missing    values.  SAS considers missing values as less than all other    values while the DBMSs don't consider them at all. */ data foo; length i 8; length x 8; length y $ 1; alpha="aAzZ09"; do i=1 to 30;   x=floor(ranuni(1)*3);   if ( ranuni(1) < 0.1 )   then x=.;   substr(y,1,1)=substr(alpha,floor(ranuni(1)*length(alpha))+1,1);   if ( ranuni(1) < 0.1 )   then y="";   output; end; drop alpha; run; proc print data=foo(where=(x < 2)); run; /*   Define library X, specifying an ACCESS engine... */ data x.foo; set foo; run; data bar; set x.foo(where=(x < 2)); run; proc sort data=bar; by i; run; proc print data=bar; run; ... View more

Thank you Cynthia. I use the libname method, which seems to be the only way mentioned for accessing Excel in SAS base programming 1 and 2 training. This week I attended PugSUG meeting where Vince gave a talk about SAS and Excel. He also gave us some great examples. I will give the ODS approach a try. ... View more

Thank you so much! I went back and revised my code. It's less clumsy now. You introduced the MOD function to me :). I've signed up for the SAS SQL training so I will review your code after that study... data work.NameList; set sasuser.pilots(keep=lastname)End=last ; Length GroupList $ 500; retain GroupList; Grouplist=Catx(';',Grouplist,LastName); if mod(_n_,5)=0 then do; output; GroupList=''; end; else if last=1 then output; drop lastname; run; proc print data=work.namelist; run; ... View more

Hi; Just imagine, if you use the same variable 3 times conditionally say for eg in if-then-else construct. if a>100 then length='Long'; else if 50<=a<=100 then length='Normal'; else if a<50 then length='short'; During compilation, the first instance the variable 'length' above is coupled with the value 'Long' which is only 4 bytes long. This is the case which you are confused of. In the second case, even if the same variable is assigned with value 'Normal', which is 6 bytes long, it cannot be accommodated, means its get truncated to 4 char length. So, the first value  which is 4 char is determined as the variable's length for later use! To avoid such truncations, SAS have LENGTH  statement which you should use well before you actually use the variables in your statements like a declaration. You need to find the space needed for the longest value of the variable  and then assign this number to that. So that there is no truncation happens like I said before. Hope this helps! LENGTH length $ 6; if a>100 then length='Long'; else if 50<=a<=100 then length='Normal'; else if a<50 then length='short'; /* Now its perfect*/ Cheers! Jojan. ... View more

Thanks! Your solution works great. I have to apologize that I pasted in the wrong code yesterday. There were titles and footnotes in the code, which might be the reason. Ods csvall file='Flight.csv'; Title 'Frequency by destination and origin'; Footnote 'Update needed'; Proc freq data=sasuser.Europe; tables dest*orig /crosslist; Run; Title 'Details'; Footnote 'Listed by destination'; Proc print data=sasuser.Europe; id Dest; var flight revenue; Run; Ods csvall close; ... View more

Thank you for the details, Cynthia. I also forgot that format has nothing to do with storage so it does not hurt to define a large one. ... View more

If you want to order by the summarized AMOUNT total, then you're going to have to make 2 passes through the data one way or the other. If there is a valid report reason for needing to do this, then in my way of thinking, you'd pick the report approach that best creates the FINAL output you want. If efficiency is your major concern, then you'll have to write, benchmark and test all the possible different solutions on the same size subset of data to see which approach is most efficient. In my experience, however, "report cosmetics" or "ease of code maintenance" can overcome efficiency in picking an approach. cynthia ... View more

Thanks to all three of you! I did not know we could combine While/Until with individual index variable values, multiple times ...in the do statement. Tested a few. I have to agree with Data _null_ that it is powerful. A nice surprise! ... View more

Great idea. This is the shortcut I was looking for. Thank you. ... View more