Re: bug in covariance.c/categoricals.c

[John Darrington]

On Thu, Jun 16, 2011 at 09:49:01AM -0400, Jason Stover wrote:
     On Sun, Jun 12, 2011 at 08:42:23PM +0000, John Darrington wrote:
     > You're right.  It's a problem of semantics.  The current implementation
     > encodes each categorical variables into N columns, where N is the number
     > of distinct values.  For the covariance matrix we require N - 1 columns.
     > 
     > Can you try the attached patch.  I think it will solve your problem.
     > (It causes all the ONEWAY tests to fail but we can think about that 
later).
     
     That worked.
     
Great!  We'll have to think of something better for a long term fix.  But if 
you can work with it for now, we'll decide on the best approach when necessary.

J'



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

signature.asc
Description: Digital signature