Re: NaN and missing values

[heberf]

In SAS if you perform any mathematical operation on a missing value it returns 
a 
missing value and generates a warning (this is why it seems similar to  NaN).  
Of course SAS is not a matrix language so it really only deals with scalar 
operations.  Following this rule makes sense for all the element by element 
operations (+,-,.*,./,exp,log,etc) but if you stuck to this rule for matrix 
operations then matrix multiplation would put a missing value in the result 
matrix whenever there is any missing value in the corrosponding row or column 
of 
the two matrices being multiplied.  Matrix division (equation solving) makes no 
sense at all.  For stats functions you could drop rows as I said before and 
then 
do normal matrix operations.

Heber

> What functions make sense for missing data?  Stats functions, I can
> understand, but I don't think it makes sense to add a matrix with
> missing data to a full matrix, for example.  Adding to matrices with
> data missing in the same locations might make sense though.  What do
> other systems (SAS, SPSS, S-plus, R, etc.) allow?
> 
> jwe
> 
> 
> 
> ---------------------------------------------------------------------
> Octave is freely available under the terms of the GNU GPL.  To ensure
> that development continues, see www.che.wisc.edu/octave/giftform.html
> Instructions for unsubscribing: www.che.wisc.edu/octave/archive.html
> ---------------------------------------------------------------------
> 



---------------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.  To ensure
that development continues, see www.che.wisc.edu/octave/giftform.html
Instructions for unsubscribing: www.che.wisc.edu/octave/archive.html
---------------------------------------------------------------------