topic Dropping a variable from a dataset conditionally in SAS Programming

Dropping a variable from a dataset conditionally

meighanj — Mon, 03 Dec 2018 11:03:38 GMT

I am trying to conditionally drop variables from a data set.

%macro drop_correlated(data_set);
%macro dummy; %mend dummy;

	proc corr data=&data_set. pearson NOPROB nomiss Rank NoProb outp=correlations;                                           
	run;

	data reduced;
		set correlations;
		where _TYPE_ = "CORR";
	run;

	data reduced;
		set reduced;
		drop _TYPE_;
	run;

	data _null_;
		set reduced;
		call symput(cat('VAR_',_N_),compress(trim(left(_NAME_))));
		call symput('NUM_VAR',_N_);
	run;
		
	%local i;
	%do i=1 %to &NUM_VAR.;

		data temp_&i.;
			set reduced;
			where _name_ = "&&VAR_&i.";
			drop _name_;
		run;

		%local j;
		%do j=1 %to &NUM_VAR.;

			data temp_&i.;
				set temp_&i.;
				if (-0.5 <= &&VAR_&j. <= 0.5) then %drop_var(temp_&i, &&VAR_&j.);
			run;
		%end;
	%end;
	%put _ALL_;

%mend drop_correlated;

%macro drop_var(data_in, var);
%macro dummy; %mend dummy;

	data &data_in.;
		set &data_in.;
		drop &var.;
	run;

%mend drop_var;


%drop_correlated(data);

I get the error:

ERROR 180-322: Statement is not valid or it is used out of proper order.

I think this is coming from the drop_var macro.

Re: Dropping a variable from a dataset conditionally

WarrenKuhfeld — Mon, 03 Dec 2018 11:21:39 GMT

You are trying to call a data step conditionally from a data step. You can't do that.

Re: Dropping a variable from a dataset conditionally

Kurt_Bremser — Mon, 03 Dec 2018 11:32:37 GMT

What you want to do is nonsense, your code would be equivalent to

data temp;
set temp;
if (-0.5 <= var <= 0.5) then drop var;
run;

Drop is a non-conditional statement. You can't drop a variable for a specific row, only for the whole dataset.

Re: Dropping a variable from a dataset conditionally

meighanj — Mon, 03 Dec 2018 11:34:06 GMT

The data set only has one row though.

Re: Dropping a variable from a dataset conditionally

RW9 — Mon, 03 Dec 2018 11:47:23 GMT

Please post example test data in the form of a datastep:

https://communities.sas.com/t5/SAS-Communities-Library/How-to-create-a-data-step-version-of-your-data-AKA-generate/ta-p/258712

And show what you want out at the end. i would guess a data _null_ to create the list then a drop would be best, something like (assumes missing is to be dropped):

data _null_;
  length dlist $2000;
  set have;
  array c{*} _character_;
  array n{*} _numeric_;
  do over c;
    if missing(c) then dlist=catx(" ",dlist,c);
  end;
  do over n;
    if missing(n) then dlist=catx(" ",dlist,n);
  end;
  call symputx('dlist',dlist);
run;
data want;
  set want (drop=&dlist.);
run;