Date:         Wed, 2 Feb 2000 11:25:34 +0100
Reply-To:     detecsm_hellriegelg@WESTLB.DE
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Gehard Hellriegel <detecsm_hellriegelg@WESTLB.DE>
Subject:      Re: delete many variables
Content-type: text/plain; charset=us-ascii

Yes, you are right! I changed this and used the ABS - function! See my other
mail - the first was too fast. Also some other suggestions where interesting,
e.g. to use mean and std. Also min=max=0 could work.








F a b r i z i o <Fabrizio1@USA.NET> on 02.02.2000 10:48:33

Bitte antworten an F a b r i z i o <Fabrizio1@USA.NET>

An:   SAS-L@LISTSERV.UGA.EDU
Kopie:     (Blindkopie: DeTeCSM HellriegelG/D/ExternalStaff/WLB)
Thema:    Re: delete many variables



Gehard Hellriegel wrote in message <41256878.00587937.00@data.WestLB.de>...
>I see no other chance than to read the whole dataset in one step to find ot
the
>zero-variables. Comfortable but not the fastest is PROC SUMMARY:
>
>%let yourdata=YOUR.DATA;
>proc summary data=&yourdata;
>var c1 c2 c3 c4 c5 ......;  /* list all your numeric variables c1-c?? */
>output out=test(drop=_type_ _freq_) sum=;
>run;


This is a dangerous one if your data might contain negative
values. Consider you have values -3, -1 and +4 in one col,
then your PROC SUMMARY will add up to zero, but the
column does not just contain zeroes.

You might want to calculate moth the MIN=, MAX=, and NMISS=.

Brgds; Fabrizio





      Mit freundlichen Gruessen


      DeTeCSM, Gerhard Hellriegel

      WestLB
      Abteilung: 001-80622
      Aderstr. 22
      D - 40217 Duesseldorf
      Tel.: +49211 826 6173
      Fax: +49211 826 5393