Testing fisher

[David Relson]

pi,

Let me summarize the result with a short table:

algorithm    min_def    spam_cutoff    test.spam    test.ham
                                        right  F-N   right  F-P
fisher          0.25          0.60      4335  191   15362    0

For comparison purposes, could you fill in your earlier test results?

David

At 08:51 AM 1/28/03, Boris 'pi' Piwinger wrote:

>Greg Louis wrote:
>
> > Looks like min_dev of 0.025 and cutoff 0.6 worked quite well for you
> > with the robx and robs values you employ.  That's interesting; I was
> > doing some tuning yesterday and found that 0.25 worked best for me
>
>I did also test this in the meantime:
>
>algorithm=fisher
>min_dev=0.25
>spam_cutoff = 0.60
>Spam:
>    4335 test.spam
>False negatives:
>191
>Ham:
>   15362 test.ham
>False positives:
>0
>
>So it gives me a lot more false negatives.