topic Re: Keep/Drop variables based on one variable in SAS Programming

Keep/Drop variables based on one variable

Xiaoningdemao — Mon, 18 Jul 2016 18:28:24 GMT

Dear All,

I have a question about drop / keep variables. Now I have such a data set:

age gender height keep drop

10 1 160 age gender

11 2 159 * height

8 1 161 * *

So the first three variables are the observation info, and the last two variables listed the variables that I want to kee/drop. So the desired fresult would be:

age

I can not use keep or drop because the actual data set contains almost 1000 variables, it is unrealistic to list them. I have been struggled with this for almost two days.

Thank you in advance!!

Best,

Thank

Re: Keep/Drop variables based on one variable

PGStats — Mon, 18 Jul 2016 18:56:43 GMT

No wonder you have been struggling. The first observation says keep age and drop gender. So where does height go?

The second observation: It says keep nothing and drop height. Yet age is kept. And where does gender end up?

How is this supposed to work? How will you know what the values kept actually are?

Re: Keep/Drop variables based on one variable

Xiaoningdemao — Mon, 18 Jul 2016 19:17:10 GMT

Hi PG,
I meant the variable 'keep' stores the variable names I want to keep, 'drop' stores the variables I want to drop, we only need to use one of it. But i don't know which is easier to use. @PG Stats@PG Stats

Re: Keep/Drop variables based on one variable

Xiaoningdemao — Mon, 18 Jul 2016 19:37:51 GMT

And since the 1000 variables we have here has both numerical and character variables I cannot simply transpose it.

Re: Keep/Drop variables based on one variable

ballardw — Mon, 18 Jul 2016 19:47:58 GMT

Keep and Drop are not conditional. They affect the entire structure of a data set. A variable that is dropped is removed entirely from the dataset. If you use Keep, then only the referenced data is kept. If you use more than one Keep then all variables that appear on any keep statement are in the final data:

Keep age;

Keep height;

means the only variables in the result set will be age and height.

And from the online help for the Keep statement:

Note: Do not use both the KEEP and DROP statements within the same DATA step.

It may be easier to use WHERE clauses in other steps to select records with desired properties. With 1000 variables in your data that may be difficult.

Re: Keep/Drop variables based on one variable

Reeza — Mon, 18 Jul 2016 19:50:46 GMT

This is a really weird data structure.

I I would use a previous step to pull names into a macro variable and execute it. I do wonder if going back a few steps would help this process.

Proc SQL noprint;
Select keep into :keep_list separated by ' '
From have;
Quit;


Data want;
Set have;

Keep &keep_list;
Run;

Re: Keep/Drop variables based on one variable

Reeza — Mon, 18 Jul 2016 19:51:51 GMT

Your also aware of variable shortcut lists I assume?

http://support.sas.com/documentation/cdl/en/lrcon/68089/HTML/default/viewer.htm#p0wphcpsfgx6o7n1sjtqzizp1n39.htm

Re: Keep/Drop variables based on one variable

Xiaoningdemao — Mon, 18 Jul 2016 20:02:28 GMT

Dear Reeza,
Thank you very much! This is exactly what I want!!
Best,

Re: Keep/Drop variables based on one variable

Xiaoningdemao — Mon, 18 Jul 2016 20:04:30 GMT

hi dallardw,
But I did put keep and drop in one data step before, i didn't recall any weird result though.... O_O

Best,

Re: Keep/Drop variables based on one variable

Xiaoningdemao — Mon, 18 Jul 2016 21:45:46 GMT

No, I didn't. But this is very helpful as well! Thanks again.

Best,