Labelling my data with labels corrected in a CSV file.

[Ninna1]

Hello fellow programmers!

I have tried to figure this out myself, but I guess I need help:

I have a large amount of survey data, and all the answer-variables have labels that are the survey questions - naturally. Not all of them are correct from the beginning, so I put them in a csv file and send them to be reviewed. Usually there is not anything to correct, so I just leave it there. If there is a single mistake i just correct it manually in my programming.

But now, there is many variables that needs new labels, so I would like to be able to do it more clever than copy/paste.

The csv file is just two collums, one with the variable names and one with the labels.

Any ideas or tips?

Looking forward to hearing from you!

Regards

[JosvanderVelden]

/* 
Post_title:
	Labelling my data with labels corrected in a CSV file.
Link to post: 
	https://communities.sas.com/t5/SAS-Programming/Labelling-my-data-with-labels-corrected-in-a-CSV-file/m-p/775060
Important post details:	
	The csv file is just two collums, one with the variable names and one with the labels.

*/

*if you are using SAS OnDemand for Academics (ODA) 
    change your-userid-on-oda for your userid
 if you are using another option then change
    /home/your-userid-on-oda/ for a path/folder 
    you can use;
%let mypath = /home/your-userid-on-oda/;

data work.dataset_with_labels;
   variable_name = "abc";
   variable_label = "label-for-abc";
run;

proc export data = work.dataset_with_labels
            outfile = "&mypath.dataset_with_labels.csv"
            replace;
run;

proc import datafile = "&mypath.dataset_with_labels.csv"
            out = work.imported_dataset_with_labels
            replace;
run;

Does this make sense for you?

Best regards, Jos

[Ninna1]

Yes it makes perfect sense but i can't really see how this brings me any closer to replacing be wrong labels with the correct ones?

[FreelanceReinh]

Hello @Ninna1,

So, you have

1. a dataset with variables containing the answers to survey questions and some of these variables have wrong labels
2. a CSV file with pairs of variable names from the above dataset and correct labels.

Let me create sample data for demonstration:

data have;
array s[*] $ Q1a Q1b Q2 Q3;
label Q1a='First question, part a)'
      Q1b='Wrong label'
      Q2 ='Second question'
      Q3 ='Again, incorrect label';
input s[*];
cards;
Yes No Maybe Never
; 

filename namlab 'C:\Temp\correct_labels.csv';

data _null_;
file namlab;
put 'Q1b,"First question, part b)"' /
    'Q3,Third question';
run;

To correct the labels in dataset HAVE you can create PROC DATASETS code (in a temporary file) using the information from the CSV file and then run it per %INCLUDE statement:

filename labels temp;

data _null_;
file labels;
infile namlab dlm=',' dsd end=last;
input name :$32. label :$256.;
if _n_=1 then put 'proc datasets lib=work nolist;' /
                  'modify have;' /
                  'label ';
put name '=' label;
if last then put '; quit;';
run;

%inc labels;

[Ninna1]

Thank you so much for your reply @FreelanceReinh ! I seems like it could work!

However I get an error. It doesn't seem to accept the:

put name '=' label;

I get this :

ERROR 73-322: Expecting an =.

ERROR 76-322: Syntax error, statement will be ignored.

[Kurt_Bremser]

We need to see an example of your csv file with names and labels to provide tested code that works with it.

Copy/paste this text into a box opened with this button:

[FreelanceReinh]

---

@Ninna1 wrote:

Thank you so much for your reply @FreelanceReinh ! I seems like it could work!

However I get an error. It doesn't seem to accept the:

put name '=' label;

I get this :

ERROR 73-322: Expecting an =.

ERROR 76-322: Syntax error, statement will be ignored.

---

I suspect that you are using non-standard variable names, e.g. containing blanks such as Question 1a. In this case you will have

options validvarname=any;

in effect and you need to modify the data _null_ step as follows in order to handle name literals:

data _null_;
file labels;
infile namlab dlm=',' dsd end=last;
input name :$32. label :$256.;
length nl $35; /* safer: $51 */
nl=nliteral(name);
if _n_=1 then put 'proc datasets lib=work nolist;' /
                  'modify have;' /
                  'label ';
put nl '=' label;
if last then put '; quit;';
run;

Edit: Here is a complete example using name literals:

Spoiler

options validvarname=any;

data have;
array s[*] $ 'First Q a'n 'First Q b'n 'Second Q'n '3rd Q.'n; /* name literals -- not recommended */
label 'First Q a'n = 'First question, part a)'
      'First Q b'n = 'Wrong label'
      'Second Q'n  = 'Second question'
      '3rd Q.'n    = 'Again, incorrect label';
input s[*];
cards;
Yes No Maybe Never
; 

filename namlab 'C:\Temp\correct_labels.csv';

data _null_;
file namlab;
put 'First Q b,"First question, part b)"' /
    '3rd Q.,Third question';
run;

filename labels temp;

data _null_;
file labels;
infile namlab dlm=',' dsd end=last;
input name :$32. label :$256.;
length nl $35;
nl=nliteral(name);
if _n_=1 then put 'proc datasets lib=work nolist;' /
                  'modify have;' /
                  'label ';
put nl '=' label;
if last then put '; quit;';
run;

%inc labels;

[Ninna1]

Thank you very much for your effort @FreelanceReinhard. It almost works.

I get a ; instead of the =

I have no idea why. However, I have allready spent too much time on this today and I have a deadline. So I will have to work on it some other time.

Thank you all.

[FreelanceReinh]

---

@Ninna1 wrote:
I get a ; instead of the =

I have no idea why. However, I have allready spent too much time on this today and I have a deadline. So I will have to work on it some other time.

---

No problem. When you get back to this, please post the complete log of the step that doesn't work into a code box opened with the </> ("Insert Code") button.

[Tom]

Make sure to define nl long enough for any possible nliteral.  You need 32 characters for the actual name plus three for the quotes and N suffix.  But you also need up to 16 more for doubling up any included quotes.  So use

length nl $51;

[FreelanceReinh]

Thanks, @Tom, good point. Indeed, the "worst case" regarding variable names that I had envisioned was not bad enough.