On 22.04.2012 07:22, Steve Fatula wrote:

    *From:* Stevan Bajić <ste...@bajic.ch>
    *To:* "dspam-user@lists.sourceforge.net"
    <dspam-user@lists.sourceforge.net>
    *Sent:* Friday, April 20, 2012 11:39 AM
    *Subject:* Re: [Dspam-user] Increase Spam Hit Rate


    3) Now go on and train with dspam_train: dspam_train SpamHitRate
    [spam_corpus maildir or mbox] [nonspam_corpus maildir or mbox]

After training with the provided Spam corpus, and, my own, and my own HAM, I get this:

SpamGlobalMerge:
TP True Positives:                599644
TN True Negatives:                 33521
FP False Positives:                   70
FN False Negatives:                 5629
SC Spam Corpusfed:                     0
NC Nonspam Corpusfed:                  0
TL Training Left:                      0
SHR Spam Hit Rate                 99.07%
HSR Ham Strike Rate:               0.21%
PPV Positive predictive value:    99.99%
OCA Overall Accuracy:             99.11%

From all the 633K messages you processed you only had 70 falsely classified as Spam while you had 5629 falsely classified as Ham. This tells me that your Ham corpus is not very diverse.

Anyway... the SHR, HSR, PPV or OCA is not important for that group.

So, looking up from what I had obviously!

Did you enable the group support for all the users and deleted their old tokens and statistics?


Looking forward to seeing what happens for the next month or so.


--
Kind Regards from Switzerland,

Stevan Bajić

------------------------------------------------------------------------------
For Developers, A Lot Can Happen In A Second.
Boundary is the first to Know...and Tell You.
Monitor Your Applications in Ultra-Fine Resolution. Try it FREE!
http://p.sf.net/sfu/Boundary-d2dvs2
_______________________________________________
Dspam-user mailing list
Dspam-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/dspam-user

Reply via email to