topic Transforming a categorical variable into a binary variable in New SAS User

Transforming a categorical variable into a binary variable

EmilyAV — Tue, 26 Oct 2021 16:55:16 GMT

Hello, I have a large dataset that contains multiple categorical variables.

For each variable, entry options are:

Yes / No / Unknown / Missing

I need to transform each variable into a binary variable: Yes vs. No/Unknown/Missing.

Two questions:

1) Is there a way to do this directly in proc freq or is it necessary to create a large number of new variables (e.g. apple_bin, orange_bin) in a data step first?

2) If a data step is necessary, is there a way to generate all of these new variables using a do loop? Each variable has a different name (e.g., apples, oranges, bananas) so I can't use a range (var1-var3).

Thanks very much,

Emily

Re: Transforming a categorical variable into a binary variable

Reeza — Tue, 26 Oct 2021 17:00:39 GMT

One method illustrated here and then several links at the bottom with other options:
https://communities.sas.com/t5/SAS-Communities-Library/How-to-create-dummy-variables-Categorical-Variables/ta-p/308484

Re: Transforming a categorical variable into a binary variable

Reeza — Tue, 26 Oct 2021 17:01:31 GMT

FYI - most procs in SAS don't require this if they support the CLASS statement which will do it as part of the PROC. If you're using PROC LOGISTIC, PHREG for example it will do this automatically, no need to recode them yourself manually.

Re: Transforming a categorical variable into a binary variable

PeterClemmensen — Tue, 26 Oct 2021 17:02:09 GMT

Should No/Unknown/Missing all transform to 0 ?

Re: Transforming a categorical variable into a binary variable

Astounding — Tue, 26 Oct 2021 17:53:33 GMT

It depends on your ultimate objective. If you just want to run PROC FREQ, that can be done easily:

proc format;
   value $yes  'Yes' = '1'    'No', 'Missing', 'Unknown' = '0';
run;

proc freq data=have;
   tables _character_;
   format _character_ $yes.;
run;