Date:         Fri, 3 Mar 2000 17:47:25 -0800
Reply-To:     Andrew James Llwellyn Cary <ajlcary@CARYCONSULTING.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Andrew James Llwellyn Cary <ajlcary@CARYCONSULTING.COM>
Subject:      Re: deleting records where all numeric variables = 0

One Solution uses a trick in SAS: a logical operator can be mixed with
arithmetic operators to build a count of non-zero values.


ARRAY _NumVars_ [] _NUMERIC_ ;

NNonZero=0;

DO __i =1 to DIM(_NumVars);
     NNonZero = NNonZero + (_NumVars[__i] ~=  0 );
    END;

IF NNonZero = 0 THEN DELETE;

DROP __i NNonZero;


This method treats missing values as non-zero.


Another way uses the max and min functions...

IF MAX(OF -NUMERIC-) = 0 AND MIN(OF _NUMERIC_) = 0 THEN DELETE;

This method ignores missing values (effectively treating them as zero).


Caveat: none of this code was tested- there may be typos...

--
Andrew James Llwellyn Cary
Sr. Curmudgeon
Cary Consulting Services
http://www.CaryConsulting.Com



"Nick Treadgold" <Nick_Treadgold@STATS.GOVT.NZ> wrote in message
news:4C256897.000C4EF6.00@mail.stats.govt.nz...
> I would like to condense a dataset by deleting all records where all
> numeric variables = 0.
>
> this is one option....
>
> if var1=0 and var2=0 and var3=0 then delete ;
>
> but i'm sure there is a simpler and more obvious solution.
>
> Note:
> this is not a suitable option for me....
>
> if sum(of var1-var3)=0 then delete;
>
>
> Nick Treadgold
> Economic Statistician
> Statistics New Zealand
> Phone: (03) 374 8834
> Fax: 03 374-8899
> Email: nick_treadgold@stats.govt.nz
> Home Page: http://www.stats.govt.nz
> The information in this email, and any files transmitted with it, is
> confidential and is for the intended recipient only.   If you receive this
> message in error, please phone us (collect) on 04-495-4600, or notify us
> via postmaster@stats.govt.nz