[Spambayes] Data file "out of balance"...?
Kenneth Sole
sole at soleassociates.com
Sun Jul 25 13:21:03 CEST 2004
Hi Tony,
When I tried to add to the DB by training only on hams, SB balked. It simply
did not occur to me that I could give it a bunch of hams and one spam, but
that worked just fine.
I sincerely appreciate your assistance,
--
Kenneth
Sole & Associates, Inc.
Box 292
Durham, New Hampshire 03824
Voice: 603-659-3169
Fax: 603-659-2248
Email: sole at soleAssociates.com
URL: http://www.soleAssociates.com
PGP: http://wwwkeys.ch.pgp.net:11371/pks/lookup?op=get&search=0xE17941C6
-----Original Message-----
From: spambayes-bounces at python.org [mailto:spambayes-bounces at python.org] On
Behalf Of Tony Meyer
Sent: Saturday, July 24, 2004 10:04 PM
To: 'Kenneth Sole'; spambayes at python.org
Subject: RE: [Spambayes] Data file "out of balance"...?
> As a result, the ratio of spams to hams in my database
> quickly goes up over 2:1 which I understand from the FAQs is
> not the best way to have things set up.
I wouldn't worry about a 2::1 (or 1::2) imbalance. Anything up to around
5::1 is probably fine - and if you keep getting the results that you want,
then don't worry about the imbalance at all.
> When this happens, I have thought to train on more hams in
> the hope of getting the DB into better "balance" but I can't
> figure out how to train on hams only. I can't move hams from
> the spam folder because none are in there.
>
> What is the best way for me to handle the situation I
> describe?
Put the messages to train in an otherwise empty folder, and use the
SpamBayes Manager dialog's Training tab to "Train Now".
=Tony Meyer
---
Please always include the list (spambayes at python.org) in your replies
(reply-all), and please don't send me personal mail about SpamBayes. This
way, you get everyone's help, and avoid a lack of replies when I'm busy.
_______________________________________________
Spambayes at python.org
http://mail.python.org/mailman/listinfo/spambayes
Check the FAQ before asking: http://spambayes.sf.net/faq.html
More information about the Spambayes
mailing list