[Spambayes] Score Interpretation
Tony Meyer
tameyer at ihug.co.nz
Thu Jan 19 09:43:36 CET 2006
> I am having trouble interpreting the score from the spam clues. The
> information is listed below. How do I tell if the mail will be
> marked as spam or not? Many thanks for your help.
>
> Combined Score: 0% (1.32361e-005)
> Internal ham score (*H*): 0.999974
> Internal spam score (*S*): 2.10942e-015
>
> # ham trained on: 1
> # spam trained on: 0
In addition to Jesse's comments: notice (if you haven't already) that
you have only trained on one ham and no spam. This means that
*everything* will look like ham (because the classifier doesn't know
what anything else looks like). You won't really get results that
make a lot of sense until you've trained on at least 5 messages of
each type.
Note also that it's generally a good idea to keep the number of ham
and spam trained roughly balanced, if possible. 2::1 or 1::3 isn't
really a problem, but you may get very odd results with 50::1 or
1::200 etc.
=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.
http://www.massey.ac.nz/~tameyer/writing/reply_all.html explains this.
More information about the SpamBayes
mailing list