<snip>
>
> Dave,
>
> I have read that 50% of alcholoic beverage sales are to alcoholics.
> That seems reasonable. 60% of adults drink and 6% of adults are
> alcoholic. That would mean that the typical alcoholic drinks 10 times
> more than the typical drinker who is not addicted.
>
> A lot of data is available related to alcoholism: suicides, murders,
> emergency room admissions, liver chirosis as a cause of death, drunk
> driving arrests and accidents, prescriptions, persons in rehab, etc.
>
> Is Anheuser-Busch including variables to account for the number of
> alcoholics living near each store?
>
> Bob
>
> 251026473393872
Bob , No need to as the level of historical level of demand would
already take that into account. The only time that you might consider
incorporating these kinds of variables ( which by the way would have
to be predicted themselves ) is iff you felt that there were dynamics
in these series that would change demand ( like a busload of
alcoholics due to stop near a store ) that would have an effect on
demand.
In fact if one or more of these variables had historically changed and
consequently had an historical effect on demand BUT was omitted from
the model , the ARIMA structure proxies such omitted variables. The
ARIMA proxy is quite appropriate unless future values of the omitted
cause variables are significantly different than what would have been
anticipated by THEIR PAST .
Regards
Dave Reilly
P.S. AB recently released actual results from the store level
forecsting by day by store by SKU activities.
Actual Benefits of Solution implemented
� Increased revenue significantly (above expectations) by reducing
out-of-stock conditions at product retailers across the United
States by
51%.
� Improved product freshness by more than � -a key marketing goal of
A-B.
� Reduced cost and increased customer satisfaction by reducing
excess
inventory on retailer premises by .1 % .
� Increased sales by a whopping 5.8% while deliveries remained
unchanged.
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
