Outlier Detection?

[Warren Togami <wt...@redhat.com>]

http://ruleqa.spamassassin.org/20090820-r806090-n/FM_FRM_RN_L_BRACK/detail

Is there an automated way to find statistical outliers like this, where 
folders known to be a foreign language trigger far more than the average?

This might help us to identify other rules that have bugs.  It is 
possible some rules like this need not be adjusted downward by the GA.

Warren Togami
wtogami@redhat.com

Re: Outlier Detection?

[Justin Mason <jm...@gmail.com>]

We don't currently have any way to spot these - I agree it'd be nice :)

--j.

Sent from my iPhone

On 21 Aug 2009, at 05:03, Warren Togami <wt...@redhat.com> wrote:

> On 08/20/2009 09:24 PM, Warren Togami wrote:
>> http://ruleqa.spamassassin.org/20090820-r806090-n/FM_FRM_RN_L_BRACK/detail
>>
>> Is there an automated way to find statistical outliers like this,  
>> where
>> folders known to be a foreign language trigger far more than the  
>> average?
>>
>> This might help us to identify other rules that have bugs. It is
>> possible some rules like this need not be adjusted downward by the  
>> GA.
>
> http://ruleqa.spamassassin.org/20090820-r806090-n/OBFUSCATING_COMMENT/detail
>
> Here is another rule abnormally triggering on Japanese only.
>
> Warren
>

Re: Outlier Detection?

[Warren Togami <wt...@redhat.com>]

On 08/20/2009 09:24 PM, Warren Togami wrote:
> http://ruleqa.spamassassin.org/20090820-r806090-n/FM_FRM_RN_L_BRACK/detail
>
> Is there an automated way to find statistical outliers like this, where
> folders known to be a foreign language trigger far more than the average?
>
> This might help us to identify other rules that have bugs. It is
> possible some rules like this need not be adjusted downward by the GA.

http://ruleqa.spamassassin.org/20090820-r806090-n/OBFUSCATING_COMMENT/detail

Here is another rule abnormally triggering on Japanese only.

Warren