topic Re: Maximum lenght of variables in SAS Programming

Maximum lenght of variables

bayzid — Wed, 03 May 2023 20:54:06 GMT

How can i find out maximum length of all the variables in a SAS dataset and then use those lengths to create a new dataset?

Re: Maximum lenght of variables

Astounding — Wed, 03 May 2023 20:57:02 GMT

You don't need to. This program will create a new data set with zero observations, but using all variable definitions from the old data set (including lengths):

data new_dataset;
   stop;
   set old_dataset;
run;

If you need a printed version of the data set structure, use PROC CONTENTS.

Re: Maximum lenght of variables

data_null__ — Wed, 03 May 2023 21:03:48 GMT

Do you mean you want to resize the character variables to the length of the longest value?

Re: Maximum lenght of variables

bayzid — Wed, 03 May 2023 23:11:47 GMT

Yes- I want to do that for both numeric and character variables. Some of the variables have greater length than maximum length. Using COMPRESS option in the dataset can reduce the size but do not change the variable length which creates an inflated dataset when transformed to other formats.

Re: Maximum lenght of variables

AhmedAl_Attar — Wed, 03 May 2023 23:25:25 GMT

Hi @bayzid

You can use my attached macro (etl_shrinkMyData.sas) It does exactly what you are looking for.

%include "etl_shrinkmydata.sas"; 
%etl_shrinkMyData(p_inDsName=SASHELP.PRDSALE
, p_outDsName=WORK.shrunk, p_noCompress=);

Hope this helps

Re: Maximum lenght of variables

bayzid — Wed, 03 May 2023 23:36:29 GMT

Hi Ahmed,
Thanks for the macro. It worked to reduce the variable length but the size of the dataset has been inflated >4 times.

Re: Maximum lenght of variables

AhmedAl_Attar — Wed, 03 May 2023 23:40:53 GMT

@bayzid

That's most likely because your original data set was compressed while the newly shrunk data set has shorter record length.

You can run Proc Content against both data sets and compare the results.

If you compressed the shrunk data set, it will probably end up same size as the original if not smaller

Re: Maximum lenght of variables

bayzid — Wed, 03 May 2023 23:44:28 GMT

Yes- that worked. After compressing the size is smaller than the original dataset. Is there any way to include the compress option in the macro so that it doesn't require an additional data step?

Re: Maximum lenght of variables

AhmedAl_Attar — Thu, 04 May 2023 00:22:23 GMT

Use
Options compress=yes;
before calling the macro

Re: Maximum lenght of variables

Tom — Thu, 04 May 2023 02:42:43 GMT

@bayzid wrote:

Yes- that worked. After compressing the size is smaller than the original dataset. Is there any way to include the compress option in the macro so that it doesn't require an additional data step?

I didn't look at the source code of the macro, since it was shared as an attached instead of in the body of the message, but from the sample call you should be able to just add the COMPRESS= option to the output dataset name.

%etl_shrinkMyData
(p_inDsName=SASHELP.PRDSALE
,p_outDsName=WORK.shrunk(compress=yes)
,p_noCompress=
);

Or set the system option COMPRESS before calling the macro.

Re: Maximum lenght of variables

bayzid — Thu, 04 May 2023 02:44:55 GMT

Yes- that worked.