Hi All,

  I have Mysql records (timestamp) that represent when a document has 
shifted flows in a workflow system. Access to documents are restricted 
based on what flow the document currently is and what user group is trying 
to access it. User groups are a combination of usernames and actions. 
Actions are mostly "read, write, modify, delete". [1]

  So far, so good. I need to find what is the average time that a flow is 
completed and predict how long a document will take to reach the next flow.

  But each user group access the system on different times of the day, even 
different days of the week. Usually, users of an user group will access the 
document at the same time (for example, 8:00 - 17:00 on weekdays), but some 
groups with several users might 8:00 to 22:00 or even work on weekends.

  My clients system admins will be free to create users, groups and 
workflows, so I can't hard-code that a flow usually will start at a value 
and end in another[2] to find the average time or to predict when the next 
one will start.

  My best try is to find the Mysql timestamps of when a flow starts and end 
most of the time.[3]

  So I am looking for a way to discover what values start and finish 90% of 
a paricular set of flow records, and work with these values between flows.

  I've tried to use some statistics formulas:

90% = 1.65 [4]

1.65 = ($start - AverageValue) / StandardDeviation
Which translates to:
$start = AverageValue + (1.65 * StandardDeviation);

  But $start always comes up with values not present on the Mysql database, 
because timestamps can have erratic values between each other. And a friend 
warned me that if an user group that usually works between 8 - 17 has an 
user that works 15 - 23, the StardardDeviation will not work as predicted, 
having erratic values too.

  So I am open to any suggestions, since I've lost my afternoon on this 
problem.

  Any advice is appreciated, and sorry for my english (I am from Brazil).

    Many thanks,

-- 
Julio Nobrega
http://www.inerciasensorial.com.br

[1] = Sorry if the intro was too long.
[2] - And I can't ask my clients to do it for me :) It would be easier, but 
I can't.
[3] Is it really? I am really open to new ideas.
[4] = According to http://caripi.cultura.ufpa.br/dicas/biome/tabe/biotaz.htm

-- 
PHP General Mailing List (http://www.php.net/)
To unsubscribe, visit: http://www.php.net/unsub.php

Reply via email to