[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Statistics on nm.debian.org don't add up

On Thu, Jan 25, 2001 at 05:59:43PM -0500, David Z. Maze wrote:
> The "aggregate statistics" section on the nm.debian.org front page
> don't make any sense.  I've found the times in the "Awaiting DAM
> Approval" line questionable, at best (every time I've looked in the
> past week or so, the mode waiting time has been 2 days, and an average 
> of 29 just doesn't seem right either).  More noticably and concretely, 
Mode is a problem because it is difficult to measure it accurately.
What happens if there are two maximums, or three.
The answer for this page at least is it uses the smallest.
You'll find a skew between mode and average due to most people going
through the process quickly, however there are those that wait a long

> the current numbers for "Waiting for AM to Confirm" are impossible:
> Count: 7
> Mode: 0
> Median: 0
> Mean: 2
> Maximum: 100
> So this means that more entries are 0 than any other value, and that
> in a sorted list 0 is the middle value.  (So at least four people have 
> been waiting zero days here.)  The mean means that the sum of all of
> the wait times is ~14 days, modulo rounding error.  This in turn means 
> that it's impossible for any single wait time to be 100 days.  (Unless 
> there's a negative wait time involved, which would solve the
> statistical mystery but seems very odd in its own right.)
"count" is what is happening right now. Right
now there are 7 people in the queue.
The wait times are based upon what has happened in the last two weeks.
In the last two weeks most (mode) people have wait 0 days.  It looks
like a very small number (mean close to 0) have waited longer.
We had a lengthy discussion previous about this and it doesn't make
sense to base statistics on what is happening in the queue at that time.

> (If the set of wait times was {0, 0, 0, 0, 0, 0, 100}, it would meet
It's more like {0,0,0...........100}. Most people accept and contact
the applicant on the same day, which is what mode tells you.

Craig Small VK2XLZ  GnuPG:1C1B D893 1418 2AF4 45EE  95CB C76C E5AC 12CA DFA5
Eye-Net Consulting http://www.eye-net.com.au/        <csmall@eye-net.com.au>
MIEEE <csmall@ieee.org>                 Debian developer <csmall@debian.org>

Reply to: