Bogotune results
Tamer Yousef
tamer.yousef at gmail.com
Wed Sep 4 18:43:38 CEST 2013
any comments about the results I got?
On Wed, Aug 21, 2013 at 12:38 PM, Tamer Yousef <tamer.yousef at gmail.com>wrote:
> I was able to finally get bogotune to run, and here are the results below.
> Here are some questions that I have:
> 1- The warning message indicates that the training set needs to
> be-classified?
> 2- applying the recommendation without the "sp_esf&ns_esf" values is
> totally screwing up the spam scores a lot of the text that previously got
> scores below .5 is now over .9.
>
> Changing the value of the min_div affects that final results significantly
> but the warning message bogotune is outputting is really making me doubt
> the while thing that I may need to re-annotate and rebuild my training
> set....
>
>
> and a side note:For applying the sp_esf&ns_esf , the "-E" option is not
> supported by recent bogfilter?1
>
>
> wordlist's ham to spam ratio is 1.2 to 1.0
> Warning: test messages include many high scoring nonspam.
> You may wish to reclassify them and rerun.
> high ham scores:
> 1 1.000000
> 2 1.000000
> 3 1.000000
> 4 1.000000
> 5 1.000000
> 6 1.000000
> 7 1.000000
> 8 1.000000
> 9 1.000000
> 10 1.000000
> low spam scores:
> 1 0.000013
> 2 0.000043
> 3 0.007088
> 4 0.029865
> 5 0.040703
> 6 0.046916
> 7 0.054538
>
>
> Minimum found at s 0.3162, md 0.286, x 0.528, spesf 0.004228, nsesf
> 0.011573
> fp 30 (2.8708%), fn 644 (62.2824%)
>
> Performing final scoring:
> Spam... Non-Spam...
> 0.254923 0.941857
> 0.298609 0.936104
> 0.307635 0.927185
> 0.362183 0.899385
> 0.413051 0.895696
> 0.466462 0.895564
> 0.470619 0.892552
> 0.471655 0.892334
> 0.472590 0.887125
> 0.477892 0.884211
>
> Recommendations:
>
> ---cut---
> db_cachesize=100
> robs=0.3162
> min_dev=0.286
> robx=0.527809
> sp_esf=0.004228
> ns_esf=0.011573
> spam_cutoff=0.936104 # for 0.10% fp (1); expect 99.32% fn (1027).
> #spam_cutoff=0.927185 # for 0.20% fp (2); expect 98.94% fn (1023).
> ham_cutoff=0.308
> ---cut---
>
More information about the Bogofilter
mailing list