Re: [AGA-mcm] Fwd: Re: What's cookin in Indiana

[Erik Olson <erik@thekrib.com>]

On Tue, 28 Jun 2005, S. Hieber wrote:

> Don't mistake me. I wasn't criticizing the components, and
> I was using "database" in a loose vernacular, like "info
> system" "computer info thing.". If I say I would like to
> see the sedan in blue, it doesn't mean I think the red one
> was not designed satisfactorily.

Actually it was an ugly shade of baby-poo yellow before and now it's 
silvery-white-ish.

> Last time we discussed this sort of thing, I came away with
> the impression that info was recorded for diff purposes
> than what I was fishing for -- tenure, rollover, maybe some
> geographics if not demographics.

Primary purpose is for generation of the mailing labels.

But in addition, I added some fields that are set by the label-generating 
script, so that we know exactly when inactive members are inactivated 
(rounded to nearest quarter).  And we also have a field for "First Joined" 
that gets set when the record is first added.  So from this we can 
directly get stats like "For all inactive members, how long between when 
they joined and when they were inactivated", and "For all active members, 
how long since they first joined?".  Neither calculation takes into 
account things like members who join, then lapse, then get re-activated. 
But still, it can be used.  And it's easy to generate (like I'll probably 
make a graph today).

In addition to the addition, since I took over the technical management of 
the database, we also have a complete record of every mailing "label" 
generated for each TAG (so far about 2.5 years).  So it's THEORETICALLY 
possible to reconstruct a more accurate picture of member behavior by 
correlating the same member from issue to issue (but a lot harder to do 
than the other).

> Basically, what I envision is a list of everyone that has
> joined AGA and their name never comes off the list -- it's
> a lasting universal truth that they joined AGA at some time
> or other.

This we have.


> For each name there is a list of the months that
> they have been members, regardless of whether the months
> are consecutive.

This we "sort-of have", in that we have the mailing label records starting 
March 2003.

> This seems to me a diff thing than a list
> of current members and how long they have signed up for,
> which is what I thought the current info system recorded --

Nope, that's NOT what we do.  We definitely do more than that.  The 
database is probably closer to what you're envisioning than what we do.

And actually, if you really think this is important, it's trivial for me 
to add another field called "Cumulative Issues Mailed" that gets bumped 
by the script every time we generate a mailing label list (as we've 
already got the logic that decrements the "Issues Remaining" field and 
inactivates non-life members when it hits zero).  I can probably even rig 
something to backfill the last 2.5 years of mailing labels into it.  Hmm.. 
just need to have some really boring long meetings to do this during...

> I'm not complaining the red is a bad color of sedan, just
> that personally, I'd like to see something in a blue. But
> I'll be the first to admit that I don't know diddly do
> about info systems. And I have no idea how much work is
> involved or whether it's worth the effort jsut to distill a
> few "facts."

Well, again, remember that there are two parts here -- the database itself 
which is just a bunch of files of data -- and the front-end WEB UI and 
scripts, which are where the real art comes into play.

  - Erik

--
Erik Olson
erik at thekrib dot com
_______________________________________________
AGA-mcm mailing list
AGA-mcm@thekrib.com
http://lists.thekrib.com/mailman/listinfo/aga-mcm