topic Re: Case insensitive in PROC MEANS in SAS Programming

Case insensitive in PROC MEANS

PeterClemmensen — Tue, 06 Dec 2016 12:04:43 GMT

Hi guys.. A very simple example below, where I create some test data such that Origin has both the character values Europe and EUROPE.

data test;
   set sashelp.cars;
   if type = 'Sedan' and Origin = 'Europe' then Origin = 'EUROPE';
run;

proc means data = test;
   class Origin;
   var Weight;
run;

When I run the proc means, can I make SAS interpret the class variable case insensitive?

Regards

Re: Case insensitive in PROC MEANS

RW9 — Tue, 06 Dec 2016 12:32:20 GMT

Well, in my opinion, this is a data modelling question rather than a coding one. What you want to end up with is a dataset contianing consistent data. Generally I like to see a dataset where textual elements contain a code as well, so in the one below you would have a step which codes all possible combinations of data element xyz to a consitent manner (be that a number or shortened code text.

So, you would code your country column based on given rules such as country in (upcase()=EUROPE, EUR, EU) all equal 1 (or EU or something).

As for your question, not sure how you would make a procedure know all the connotations of the data which only you know. As above, you state Europe=EUROPE, but what about EU which is another common abbreviation.

Re: Case insensitive in PROC MEANS

Rick_SAS — Tue, 06 Dec 2016 12:53:54 GMT

Yes. In general you can define a SAS format that collapses all of the possible choices into a single formatted value.

For your example, you don't need to create a custom format because the $UPCASE. format converts all values to uppercase:

proc means data = test;
format Origin $UPCASE.;
   class Origin;
   var Weight;
run;

Re: Case insensitive in PROC MEANS

PeterClemmensen — Wed, 21 Dec 2016 09:00:04 GMT

Thank you all for your advice and happy holidays 🙂