[vox] LUGOD meeting attendance and new-user signups

[Bill Kendrick]

For a long while, I've been curious to graph LUGOD attendance
and LUGOD member signups.[*]  I did that this afternoon.

The membership list ( http://www.lugod.org/members/ ) is driven
by a little tab-delimited data file, so it was very trivial to
figure out how many members signed up on different months.

  http://lugod.org/documents/history/attendance-histogram-200911.pdf


How I did it:

Our meeting minutes were slightly more involved; I whipped together
a PHP script to parse each of the meeting minutes HTML/PHP files,
look for "<h3>Attendance</h3>", and start counting "<li>" list items.
Any starting with "..." were grabbed literally, rather than counted,
since those are how I denote attendees whose names we didn't get
(e.g., they came in late).  "... and 5 others" or "...and two others."

I manually converted those to digits, using rough estimates.
("A few" became "2", "15-20" became "17", etc.)


A bit more difficult was figuring out how many meetings took place
on a particular month.  (Memberlist just records MM/YYYY; and we've
had anywhere between 0 and 3 meetings in a given month.)
I simply eyeballed the Past Meetings web page, and tried my best
to fill out the numbers.  This gave me an average number of new
members, per meeting, for a given month.


Finally, the hard part was assembling a pair of spreadsheets
together.  I wanted to total the attendance for each meeting
in a month (again, between 0 and 3 numbers to add up), so I can
compare "new members in month X" to "attendance for all meetings
in month X".

For those curious, the trick, in OpenOffice.org Calc, was:

  =SUMIF(Attendance.$E$2:$E$218,A2,Attendance.$F$2:$F$218)

...where "E2"-"E218" on the "Attendance" sheet is the the date
of the minute, sans the day.  I resized that column into
nothingness, but it's just, e.g.:

  =DATE(YEAR(A2),MONTH(A2),1)

...and where "F2"-"F218" of "Attendance" is the estimated
total number of attendees.  (The true number, from the B column --
derived from counting all "<li>" that appear after "<h3>Attendance"
and aren't "<li>..." -- plus the number, often an estimate, of
how many other people showed up.)


In some cases, attendance wasn't taken at ALL (or the meeting
minutes never surfaced from whoever the secretary was at the
time).  You'll notice "unknown" flags in a column in the
"Attendance" sheet.  Nothing much I can do about those, but
I use to get what I call the "real average" attendance of
our meetings -- that is, the avg. attendance for all meetings
where we have a number:

  20.5829145728643

I'm not sure who that 29/50ths of a person is who's been showing up,
but I'll keep an eye out for them.  ;)


I didn't do too much cleverly cleanup of the list of new members,
since that number is so wildly unreliable (see [*], below).
However, our greatest influx (people who signed the form) was 20,
back in July 2002.  (Hans Reiser/ResierFS and Rasmus Lerdorf/PHP
were the two speakers.)

Anyway... I'm not very good at this, but people might find the
results interesting.  Enjoy!


[*] People who come to meetings regularly, or participate in the
    club via mailing lists or IRC, often never get around to
    filling out a membership form.  Conversely, I'm sure there are
    many people who _did_ fill out a membership form when they came
    to their first LUGOD meeting, but have never come back.

-- 
-bill!
Sent from my computer