[spambayes-dev] removing punctuation redux

[Remi Ricard]

Hi,

This is my run on 1996 Spam and 1771 Ham.
In the spam sample there is some Spam with Subject like:
^*Br,ea*k o-ut the pla"sti`c..Th;is i`s it!!    etbdtpxdgptvb
B00ST' 'your `P`EN`I.S_ S*IZE_                - skhcfqc


You will have to interpret the result.



stds.txt -> puncts.txt
-> <stat> tested 185 hams & 200 spams against 1586 hams & 1796 spams
-> <stat> tested 155 hams & 182 spams against 1616 hams & 1814 spams
-> <stat> tested 182 hams & 217 spams against 1589 hams & 1779 spams
-> <stat> tested 178 hams & 203 spams against 1593 hams & 1793 spams
-> <stat> tested 206 hams & 208 spams against 1565 hams & 1788 spams
-> <stat> tested 177 hams & 196 spams against 1594 hams & 1800 spams
-> <stat> tested 181 hams & 176 spams against 1590 hams & 1820 spams
-> <stat> tested 176 hams & 221 spams against 1595 hams & 1775 spams
-> <stat> tested 175 hams & 206 spams against 1596 hams & 1790 spams
-> <stat> tested 156 hams & 187 spams against 1615 hams & 1809 spams
-> <stat> tested 185 hams & 200 spams against 1586 hams & 1796 spams
-> <stat> tested 155 hams & 182 spams against 1616 hams & 1814 spams
-> <stat> tested 182 hams & 217 spams against 1589 hams & 1779 spams
-> <stat> tested 178 hams & 203 spams against 1593 hams & 1793 spams
-> <stat> tested 206 hams & 208 spams against 1565 hams & 1788 spams
-> <stat> tested 177 hams & 196 spams against 1594 hams & 1800 spams
-> <stat> tested 181 hams & 176 spams against 1590 hams & 1820 spams
-> <stat> tested 176 hams & 221 spams against 1595 hams & 1775 spams
-> <stat> tested 175 hams & 206 spams against 1596 hams & 1790 spams
-> <stat> tested 156 hams & 187 spams against 1615 hams & 1809 spams

false positive percentages
    0.541  0.541  tied          
    0.000  0.000  tied          
    0.549  0.000  won   -100.00%
    0.000  0.000  tied          
    0.485  0.485  tied          
    0.000  0.000  tied          
    0.000  0.000  tied          
    0.000  0.000  tied          
    0.571  1.143  lost  +100.18%
    0.000  0.000  tied          

won   1 times
tied  8 times
lost  1 times

total unique fp went from 4 to 4 tied          
mean fp % went from 0.214685655463 to 0.216883457661 lost    +1.02%

false negative percentages
    4.500  4.500  tied          
    1.648  1.648  tied          
    2.304  2.304  tied          
    0.493  0.493  tied          
    3.365  3.365  tied          
    0.000  0.000  tied          
    1.136  1.136  tied          
    0.452  0.452  tied          
    2.913  2.913  tied          
    1.604  1.604  tied          

won   0 times
tied 10 times
lost  0 times

total unique fn went from 37 to 37 tied          
mean fn % went from 1.84162463248 to 1.84162463248 tied          

ham mean                     ham sdev
   0.70    0.71   +1.43%        7.52    7.52   +0.00%
   0.20    0.16  -20.00%        1.73    1.24  -28.32%
   0.51    0.51   +0.00%        6.72    6.61   -1.64%
   0.22    0.21   -4.55%        2.50    2.42   -3.20%
   0.64    0.65   +1.56%        7.25    7.27   +0.28%
   0.63    0.58   -7.94%        5.35    4.88   -8.79%
   0.01    0.01   +0.00%        0.06    0.06   +0.00%
   0.64    0.64   +0.00%        6.90    6.73   -2.46%
   1.10    1.10   +0.00%        9.97   10.03   +0.60%
   0.62    0.65   +4.84%        4.76    4.85   +1.89%

ham mean and sdev for all runs
   0.53    0.52   -1.89%        6.09    6.02   -1.15%

spam mean                    spam sdev
  93.67   93.62   -0.05%       21.46   21.55   +0.42%
  95.77   95.74   -0.03%       16.09   16.13   +0.25%
  95.65   95.67   +0.02%       17.11   17.03   -0.47%
  97.81   97.78   -0.03%        9.92    9.92   +0.00%
  95.35   95.31   -0.04%       17.67   17.74   +0.40%
  97.61   97.43   -0.18%       10.02   10.71   +6.89%
  98.16   98.15   -0.01%       10.85   10.88   +0.28%
  97.47   97.40   -0.07%       10.37   10.77   +3.86%
  95.08   95.02   -0.06%       18.05   18.16   +0.61%
  96.21   96.13   -0.08%       15.32   15.42   +0.65%

spam mean and sdev for all runs
  96.26   96.21   -0.05%       15.29   15.41   +0.78%

ham/spam mean difference: 95.73 95.69 -0.04


> =Tony Meyer
-- 
Remi Ricard <papaDoc at videotron.ca>