I am not sure if this is common knowledge or not, but I was curious about
the frequency each test identified spam so I ran some numbers and I got the
following. I was trying to determine if I should continue to run dcc, razor
and pyzor.
The 'mails' column is the total number of spam caught. The number under
each of the 4 tests is the percent of the total spam which was caught by
that particular test. Do these numbers seem accurate? I don't run a very
busy mail server, but have had a great deal of success with spamassassin. I
believe I have filtered 99.99% of spam. I would appreciate some feedback on
the stats I have produced and your opinions of the value of using dcc, razor
and pyzor.
Mails Bayes Dcc Razor
Pyzor
Feb 29 1339 1134 (84.69%) 929 (69.38%) 560 (41.82%) 595 (44.44%)
Feb 22 3748 3216 (85.81%) 2515 (67.10%) 1394 (37.19%) 1810 (48.29%)
Feb 15 2942 2514 (85.45%) 1980 (67.30%) 1364 (46.36%) 1394 (47.38%)
Josh
