[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]


From: John Darrington
Subject: covariance-matrix.c
Date: Wed, 9 Sep 2009 01:39:10 +0800
User-agent: Mutt/1.5.18 (2008-05-17)

I've started implementing CORRELATIONS, and it make sense to use
src/math/covariance-matrix.c but I have a number of questions
about it.

covariance_accumulate_listwise doesn't seem to be complete.
Notably, val2 is used uninitialised.   However, I don't really
see a need for this function, since it's straightforward, to
filter missing values on a listwise basis using a casefilter.

struct covariance_matrix has a member called "ssize".  I'm guessing
that ssize is a matrix containing the number of non-missing cases 
used to calculate each element of cov.  Such a matrix would be 
rather usefull to CORRELATIONS, but I don't see a method on the
interface to access ssize.   

However, looking at the code, I see instances of expressions like
"->ssize += 1.0" which worries me, because it doesn't seem to account 
for weighted cases.


1. Is covariance_accumulate_listwise actually used? If not, I think
   it's ok to delete it.

2. Can we have a way to access ssize?

3. Has covariance matrix been testing in the presence of non-unity
   caseweights?  If not, then it ought to be.

4. A few extra comments in covariance-matrix.c wouldn't go amiss.


PGP Public key ID: 1024D/2DE827B3 
fingerprint = 8797 A26D 0854 2EAB 0285  A290 8A67 719C 2DE8 27B3
See or any PGP keyserver for public key.

Attachment: signature.asc
Description: Digital signature

reply via email to

[Prev in Thread] Current Thread [Next in Thread]