I'm looking for raw classification rates,
That is
number of successful classifications / total number of email.
So if you have 100 emails classified as 10 good and 90 spam, and 8 are classified wrongly (say 5 classified as spam are actually good, and 3 classified as good are actually spam) , your accuracy rate is (100-5-3)/100* 100% = 92%
Edited by Paranoid, 06 July 2004 - 10:51 AM.