Date:         Tue, 1 Feb 2000 17:25:08 +0100
Reply-To:     BONNET Christophe <Christophe.BONNET@GEP.FR>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         BONNET Christophe <Christophe.BONNET@GEP.FR>
Subject:      Re: delete many variables
Comments: To: Hui Huang <huihuang@IUPUI.EDU>
Content-Type: text/plain; charset="iso-8859-1"

Here we are: one of my friend and I propose you what follows.
It uses some sql and some macro-langage but it seems quite
simple.
The macro %deli will do what you want
The first step uses a proc transpose, then you compute the mean and the std
If mean=std=0 then your variable is all 0's.

The proc sql puts every "null" variable in a list
Then you just have to drop these variables in the last step...



data a;
input c1 c2 c3 c4;
cards;
1  0  1  0
0  0  1  0
1  0  1  0
;
run;

%macro deli(data=);

proc transpose data=&data out=toto;

data toto;
set toto;
m=mean (of _numeric_);
s= std( of _numeric_);
if m=0;
if s=0;
keep _NAME_;
run;

proc sql;
select _name_ into :list separated by " " from toto;
quit;

data a;
set a;
drop &list;
run;

%mend deli;

proc print data=a;

%deli(data=a);

proc print data=a;



> ----------
> De :  Hui Huang[SMTP:huihuang@IUPUI.EDU]
> Répondre à :  Hui Huang
> Date :        mardi 1 février 2000 16:21
> A :   SAS-L@LISTSERV.UGA.EDU
> Objet :       delete many variables
>
> Hi all,
>
> Is there any way to delete many variables from a dataset?
> For exampl dataset AAA looks like this:
>
> c1 c2 c2 c4
> 1  0  1  0
> 0  0  1  0
> 1  0  1  0
>
> I want to delete variable c2 and c4 because every obs' value is 0
> for those variables. My point is how to delete some variables with all
> values of obs in those variables 0's, but you don't know exactly which
> variables have this property in advance, and therefore need to use some
> sentences to do the judgement work?
>
> Any responses will be appreciated.
>
> Hui
>