DM: Re: churn question; using prior class probs

[Lynd Bacon]

Other considerations might include:

(1)your data do not contain info about firms' marketing activities, 
e.g.
promotions and incentives. These are not constant over time, 
especially in
the telecom industry.
(2) there may be some kind of generic trend in customer switching 
behavior
as consumers become more aware of available options, and "learn" to 
switch.
(3) If you did not use test cases to evaluate your initial model, you 
may
be just experiencing "shrinkage" (Not of the George Constanza variety,
however.)

-lynd bacon

At 08:50 AM 4/9/98 -0400, Dr. Andrew A. Kramer wrote:
>Did you use training & testing data sets when you ran your model, or
>just use the 8000 observations to train the network? If the latter, 
>than
I'm not
>surprised at your results.
>
>Andrew A. Kramer, Ph.D.
>Future Analytics, Inc.
>
>Hallberg Rassy wrote:
>> 
...
/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\
|~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~|
| LYND BACON & ASSOCIATES, LTD.       http://www.lba.com |
| marketing and management science    mr.daemon@lba.com  |
| Homewood IL USA                     +1.708.957.0883    |
|                ---------------------                   |            
| 
| Find out about the Chicago ASA monthly speaker series  |
|         at http://www.lba.com/asa-lunch.html           |
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/