Date:         Tue, 29 Aug 2006 02:23:12 -0700
Reply-To:     green4x <Daniel.Miroshnichenko@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         green4x <Daniel.Miroshnichenko@GMAIL.COM>
Organization: http://groups.google.com
Subject:      Re: A question about how to let missing value join the computation
Comments: To: sas-l@uga.edu
Content-Type: text/plain; charset="iso-8859-1"

It's not the worse thing you should afraid of :) Try "missing" option
in your table statement, as it treats missing values as nonmissing.

But my advice - try to use a format for the predclass variable or
change missing values for specific number (like 777 or 999) - different
ways depending of further analysis.

Minze Su wrote:
> Hi all,
>
>   I have a dataset like this:
>
> obs class    predclass
>
> 1    1        1
> 2    2        2
> 3    1        .
> 4    2        .
> 5    2        1
> 6    1        2
>
>   Now I want to use proc freq to make a confusion matrix to output some
> statistic-KAPPA. But I only got the result without obs 3 and 4 because of
> the missing values. Is there any simple option to let this two
> observations go into the computation? Output table result and calculate
> myself is not good I think. Please help! Many thanks.
>
>
> Best regards,
>
> Minze Su