http://bugzilla.spamassassin.org/show_bug.cgi?id=2419
[EMAIL PROTECTED] changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|RESOLVED |REOPENED
Resolution|FIXED |
------- Additional Comments From [EMAIL PROTECTED] 2004-04-26 19:43 -------
-0.9
Good idea, but the wrong implementation and the wrong focus. As it stands,
the buckets are miscalibrated such that most of the buckets are way too small.
Here's a few from DETAILS.new, separated by user, sorted by overall hit rate:
53.679 84.1924 0.0091 1.000 1.00 0.00 BAYES_99:daf
35.488 0.9225 96.2852 0.009 0.99 0.00 BAYES_00:daf
2.409 3.6850 0.1641 0.957 0.78 0.00 BAYES_50:daf
2.293 3.5943 0.0046 0.999 0.89 0.00 BAYES_90:daf
1.327 1.5548 0.9253 0.627 0.05 0.00 BAYES_44:daf
0.816 1.2802 0.0000 1.000 0.89 0.00 BAYES_80:daf
0.674 0.9666 0.1595 0.858 0.00 0.00 BAYES_10:daf
0.669 1.0443 0.0091 0.991 0.87 0.00 BAYES_60:daf
0.634 0.9925 0.0046 0.995 0.88 0.00 BAYES_70:daf
0.334 0.5235 0.0000 1.000 0.89 0.00 BAYES_56:daf
0.160 0.0726 0.3145 0.187 0.48 0.00 BAYES_01:daf
0.081 0.0466 0.1413 0.248 0.38 0.00 BAYES_30:daf
0.079 0.0674 0.1003 0.402 0.19 0.00 BAYES_20:daf
0.058 0.0363 0.0957 0.275 0.34 0.00 BAYES_40:daf
47.855 1.9803 91.0000 0.021 0.98 0.00 BAYES_00:jm
20.678 42.6635 0.0000 1.000 0.97 0.00 BAYES_99:jm
8.239 16.9989 0.0000 1.000 0.94 0.00 BAYES_50:jm
6.706 13.8357 0.0000 1.000 0.00 0.00 BAYES_44:jm
2.564 5.2897 0.0000 1.000 0.92 0.00 BAYES_90:jm
1.185 2.4455 0.0000 1.000 0.92 0.00 BAYES_60:jm
1.018 2.0999 0.0000 1.000 0.92 0.00 BAYES_80:jm
0.857 1.7677 0.0000 1.000 0.92 0.00 BAYES_70:jm
0.548 1.1297 0.0000 1.000 0.92 0.00 BAYES_56:jm
0.309 0.6114 0.0250 0.961 0.00 0.00 BAYES_10:jm
0.309 0.5715 0.0625 0.901 0.00 0.00 BAYES_01:jm
0.238 0.4386 0.0500 0.898 0.00 0.00 BAYES_20:jm
0.219 0.4253 0.0250 0.944 0.00 0.00 BAYES_30:jm
0.180 0.3589 0.0125 0.966 0.00 0.00 BAYES_40:jm
36.403 52.8150 1.2213 0.977 0.97 0.00 BAYES_99:lan
1.074 1.4075 0.3592 0.797 0.49 0.00 BAYES_90:lan
1.005 0.0000 3.1609 0.000 0.96 0.00 BAYES_00:lan
0.937 1.2735 0.2155 0.855 0.60 0.00 BAYES_50:lan
0.526 0.2346 1.1494 0.169 0.55 0.00 BAYES_44:lan
0.366 0.0000 1.1494 0.000 0.96 0.00 BAYES_20:lan
0.343 0.4692 0.0718 0.867 0.62 0.00 BAYES_80:lan
0.320 0.4692 0.0000 1.000 0.96 0.00 BAYES_70:lan
0.229 0.3016 0.0718 0.808 0.50 0.00 BAYES_60:lan
0.137 0.0000 0.4310 0.000 0.96 0.00 BAYES_01:lan
0.114 0.1676 0.0000 1.000 0.96 0.00 BAYES_56:lan
0.069 0.0000 0.2155 0.000 0.96 0.00 BAYES_40:lan
0.069 0.0000 0.2155 0.000 0.96 0.00 BAYES_30:lan
0.069 0.0000 0.2155 0.000 0.96 0.00 BAYES_10:lan
The small bucket size decreases the score further (due to score ranges)
making the intermediate ranges weaker than they could be.
Yes, the high and low end are neglected, but I think this solution is
no good.
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.
