[Spambayes] Mining the headers
Sun Oct 27 05:40:27 2002
>> (I seem to have a bit different fp & fn results than others.)
Tim> Actually, unless things have changed dramatically, you get worse
Tim> results than everyone else combined. That still lacks an
Tim> explanation. Have you tried chi-combining yet?
Yes, now that it's the default, though I didn't make a conscious attempt to
compare it with other schemes. I got a bit behind the past several weeks,
and still don't really understand all of what's been changed lately (I don't
even pretend to be a statistician as a pick-up line in online chat rooms!).