topic Re: Array and changing all impossible values in SAS Studio

Array and changing all impossible values

smart — Mon, 03 Dec 2018 18:46:08 GMT

My Goal:

Create a temporary data set called Depression that reads the dpq_h_errorsSAS data set and changes all impossible values in the depression variables to missing. The dpq_h_errors data set is stored in the /courses/ddb976e5ba27fe300/NHANES 2013_2014 folder.

My code:

libname in '/courses/ddb976e5ba27fe300/NHANES 2013_2014';

data Depression;
input SEQN DPQ010 DP020 DP030 DP040 DP050 DP060 DP070 DP080 DP090 DP100;
array codebook [10] DPQ010-DPQ100;
do i = 1 to 10;
if codebook [i] = 10 then codebook[i] = 0;
end;

proc contents data=in.dpq_h_errors;
run;

proc print data=in.dpq_h_errors (obs=10);
run;

The error and warning I am getting:

ERROR: Too many variables defined for the dimension(s) specified for the array codebook

WARNING: The data set WORK.DEPRESSION may be incomplete. When this step was stopped there were 0 observations and 12 variables.

WARNING: Data set WORK.DEPRESSION was not replaced because this step was stopped.

Re: Array and changing all impossible values

novinosrin — Mon, 03 Dec 2018 18:51:08 GMT

Try this change double dash --

array codebook [10] DPQ010--DPQ100;

Re: Array and changing all impossible values

smart — Mon, 03 Dec 2018 19:01:00 GMT

When I submitted the array with double dash I received these errors in my log:

data Depression;

74 input SEQN DPQ010 DP020 DP030 DP040 DP050 DP060 DP070 DP080 DP090 DP100;

75 array codebook [10] DPQ010--DPQ100

76 do i = 1 to 10;

_ __

22 22

200 200

ERROR: DPQ100 does not follow DPQ010 on the list of previously defined variables.

ERROR: Too few variables defined for the dimension(s) specified for the array codebook.

ERROR 22-322: Syntax error, expecting one of the following: a name, (, -, :, ;, _ALL_, _CHARACTER_, _CHAR_, _NUMERIC_.

ERROR 200-322: The symbol is not recognized and will be ignored.

77 if codebook [i] = 10 then codebook[i] = 0;

78 end;

___

161

ERROR 161-185: No matching DO/SELECT statement.

Re: Array and changing all impossible values

novinosrin — Mon, 03 Dec 2018 19:16:09 GMT

74 input SEQN DPQ010 DP020 DP030 DP040 DP050 DP060 DP070 DP080 DP090 DP100;/*all DP series*/

75 array codebook [10] DPQ010--DPQ100 /*DPQ?????????*/

wouldn't the array statement be

array codebook [10] DPQ010--DP100 rather?

Re: Array and changing all impossible values

smart — Mon, 03 Dec 2018 19:20:14 GMT

That was a typo on my part. All should be DPQ. Fixing that still gives me the following errors:

ERROR: Too many variables defined for the dimension(s) specified for the array codebook.

ERROR 22-322: Syntax error, expecting one of the following: a name, (, -, :, ;, _ALL_, _CHARACTER_, _CHAR_, _NUMERIC_.

ERROR 200-322: The symbol is not recognized and will be ignored.

77 if codebook [i] = 10 then codebook[i] = 0;

78 end;

___

161

ERROR 161-185: No matching DO/SELECT statement.

Re: Array and changing all impossible values

novinosrin — Mon, 03 Dec 2018 19:25:21 GMT

can you post your code plz

Re: Array and changing all impossible values

r_behata — Mon, 03 Dec 2018 23:10:31 GMT

The column names referred in the array statement should be be contiguous to use the reference like DPQ010-DPQ100. In your case the columns are DPQ010 DP020 DP030 DP040 DP050 DP060 DP070 DP080 DP090 DP100 which are not contiguous. Moreover DPQ010-DPQ100 are seen as 90 elements by sas and the array is declared to have only 10 elements.

You have to expand the column list , which is simplified in this example.

data have;
input SEQN DPQ010 DP020 DP030 DP040 DP050 DP060 DP070 DP080 DP090 DP100;
cards;
1 10 11 12 13 14 15 16 17 18 19 20
run;

proc contents data=have out=temp(keep=name) noprint;
run;

Proc sql noprint;
	select name into :arr_lst separated by ' '
		from temp
		where name like 'DP%';
quit;

%put &=arr_lst;

data Depression;
	set have;
	array codebook [10] &arr_lst;

	do i = 1 to 10;
		if codebook [i] = 10 then
			codebook[i] = 0;
	end;
run;

Re: Array and changing all impossible values

mkeintz — Tue, 04 Dec 2018 21:41:03 GMT

You are doing a very roundabout way to get a list of variable names, based on the first 2 letters of those name.

This single step does what I think you want.

data depression;
  set have;
  array codebook {*} dp: ;
  do i=1 to dim(codebook);
    if codebook{i}=10 then codebook{i}=0;
  end;
run;

this uses the expression DP: to indicate all vars whose names begin with DP.
It uses an asterisk in the braces, telling sas to figure out how many elements are in the array
And it uses dim(codebook) instead of 10 as the do loop upper limit. There's no need for you to know the size in advance.

Re: Array and changing all impossible values

r_behata — Tue, 04 Dec 2018 21:48:33 GMT

Thanks for pointing out @mkeintz . Wondering how i missed that one.

Re: Array and changing all impossible values

smart — Wed, 05 Dec 2018 23:05:36 GMT

This works thank you so much!