Bob's Plugins for Spamihilator / Filter Statistics

Filter Statistics dialogFilter Statistics

These settings belong to the Filter Statistics plugin. Since this plugin is not a filter, it will not be listed in Spamihilator's Priorities window.

The Filter Statistics plugin is not directly integrated into Spamihilator, but it can be used to view and graph statistics from Spamihilator and its filter plugins. It can also be used to configure Spamihilator's statistics settings. Data is read from the spamihilator.ini and filter.log files (which are created by Spamihilator) and statistics are then computed by this plugin.

Features

  • General statistics are displayed in a dialog window after you click on the General Statistics button. Spamihilator statistics are calculated using the spamihilator.ini file. Filter statistics are calculated using the filter.log file. (see note #3 below)General Statistics dialog

  • Enable/disable the Filter Logging feature in Spamihilator. This setting must be enabled to see newly updated statistics for filters. By default, Filter Logging is enabled when the Filter Statistics plugin is installed.

  • Set/change the maximum size of the filter.log file. The default is 100kb (102,400 bytes). The maximum is 1GB, although this is not recommended.

  • Statistics are logged to a separate output text file in case you would like to use a seperate graphing application (e.g. Microsoft Excel). The default output text file is called "filterstats.log" and will be located in Spamihilator's \plugins folder, but you may choose your own file and/or folder. Any errors that are encountered while using this plugin will also be logged to the output text file.

  • Display a graph (bar chart or pie chart) and statistics of just one statistical category, or display all statistics by clicking on the All Statistics button. No graphs will be displayed when using the All Statistics button. In either case, statistics will also be written to the output text file mentioned above.

  • Reset (delete) the filter.log file by clicking on the Delete button. Doing so will allow you to "start from scratch", meaning you can see the filter statistics that have accumulated from a specific point in time (i.e. the time when the Delete button was clicked).

  • Reset (delete) Spamihilator's statistics in the spamihilator.ini file by clicking on the Reset Statistics button in the More Options dialog. Doing so will allow you to "start from scratch", meaning you can see Spamihilator's statistics that have accumulated from a specific point in time (i.e. the time when the Delete button was clicked). When you reset the statistics, a copy of them will be saved so you can restore them later if you wish.More Options dialog

  • Restore Spamihilator's statistics in the spamihilator.ini file by clicking on the Restore Statistics button in the More Options dialog. This is useful if you want to start from scratch, view the statistics over a certain period of time, and then want to restore the statistics to their original values.

  • Because of "orphaning" (see note #2 below), the starting date is not always known, so it will be represented by the first date that is found in the filter.log file and there will be two asterisks (**) next to the date in the Graph and General Statistics dialog windows. This will also be mentioned in the output log file.

  • Specify where the filter.log file is located if Spamihilator creates it in a different location than previous versions of Spamihilator. You cannot tell Spamihilator to create the file in a different location, but you can tell the Filter Statistics plugin where Spamihilator created it.

  • View the filterstats.log file by clicking on the View button. This button is located in the filter's main settings window. The button is also in the Bar and Pie chart windows and the General Statistics window.

  • Display the average number and size of spams, non-spams and unknowns per day. These are displayed in the output log file.

Statistical Categories
3-D Pie Chart dialog There are ten statistics in three different categories. They are:

  • Spam vs. Non-Spam
    • Number of Mails
    • Size of Mails

  • Filter vs. Filter
    • Number of Spam Mails
    • Number of Non-Spam Mails
    • Size of Spam Mails
    • Size of Non-Spam Mails

  • E-mail Account vs. E-mail Account   (see note #1 below)
    • Number of Spam Mails
    • Number of Non-Spam Mails
    • Size of Spam Mails
    • Size of Non-Spam Mails

Definitions
  • False Positives are Non-Spam mails that are mis-identified by Spamihilator as Spam. We want this number to be as small as possible.

  • False Negatives are Spam mails that are mis-identified by Spamihilator as Non-Spam. These will normally occur, but should decrease in rate as Spamihilator learns from your daily mails (assuming you are using the Training Area to teach Spamihilator).

  • Mails Classified is the number of mails used for training in the Training Area.

  • Spam Recognition Rate = 100 - (False Negatives * 100 / Mails Classified)

  • Non-Spam Recognition Rate = 100 - (False Positives * 100 / Mails Classified)

  • Overall Filter Efficiency = 100 - (((False Negatives + False Positives * 2) / 3) * 100 / Mails Classified)
    [Please note that false positives are more important than false negatives, thus they count twice.]
Notes
1. If you have only one e-mail account that Spamihilator monitors, the statistics in the "Account vs. Account" category may not be of any interest to you.

2. When selecting any of the "Account vs. Account" statistics, you may see one or more "(orphaned)" entries. Orphaned mails are mails where the e-mail account is not known because that information has "scrolled off the top" of the filter.log file. This occurs when the filter.log file is full and new data is appended to the bottom of the file.

3. There will be discrepancies between the Spamihilator and filter statistics. This is due to the fact that the Spamihilator statistics are calculated from "running total" counters from the time Spamihilator was installed (version 0.9.8.3 or later), whereas the filter statistics are calculated from the filter.log file which will have old data removed when new data is downloaded. This is due to the "scrolling" process mentioned in note #2 above.

To save changes in the "Filter Statistics" section of the Settings dialog, click on either the Apply or OK button.

You can configure this plugin by opening Spamihilator's Settings dialog and clicking on "Filter Statistics".