Ok. I quickly got through today's messages so if someone made this point I
may have missed it. The N-G model is a model for the population of boxcars,
it's a fleet model. It is not an attempt to model or describe a specific
train. We've discussed this on here before. We've clearly shown a specific
train won't necessarily fit the model (Mike Brock's SP boxcar train being
one that gets mentioned every time) but trains (plural) are analyzed the
data often supports the model.
If I am using the N-G model to develop a fleet (which I am) on my layout
those cars will run in various trains as a pool from which to operate.
I feel like a Mark twain reference to lies, damn lies and statistics is
Brian J. Carlson, P.E.
From: STMFC@... [mailto:STMFC@...] On Behalf Of
Aley, Jeff A
Sent: Tuesday, April 13, 2010 12:19 PM
Subject: RE: [STMFC] Re: Freight car Distribution - Larry Kline
OK, I'll give you (and Ned) another reason why peddler freights, turns, and
locals may not seem to follow the N-G model.
The statistical sample size is much smaller (meaning you will therefore
observe more variance).
The G-N model is a macroscopic model - it works well for large quantities of
shipper - consignee pairs. But as Dennis points out, for a small quantity of
shipper-consignee pairs, the "exceptions" may (will?) dominate.