|
In today''s marketing environment , because consumer
behaviors change,the technology of telecommunication improves
and costs of distribution rise, direct marketing will be more
and more important . In another way, through the use of
database marketing, direct marketers can build and manage
customer database to track the lifetime value of their
relationship with each customer . However, one of the most
important tasks of the database marketing is to facilitate the
creation of statistical model and the subsequent
implementationof model results. Traditional method of scoring
in direct marketing estimate recency, frequency and monetary
measures on an individual-by-individual basis which doesnot
fully utilize information that may exist across customers ,and
the RFM method does not have the ability to predict the
responses of customerspreviously. The regression model has the
ability to forecast some future event,such as the average or
overall response rate to an upcoming mailing,but it can not
forecast each individual''s response. In order to forecast each
individual''sresponse, I employ a hierarchical bayes poisson
regression model which does not
only consider the difference of individual''s purchase rate but
also consider the difference of individual''s regression
coefficient. For this reason,even independent variables are the
same,individual''s purchase rate(dependent variable)will not be
the same.
The objective of the research : (1) . Establish a model which
can predict individual customer value exactly.(2) . use a set
of real data to prove our model has the better ability of
forecast than the traditional regression model .
The result of the research: (1) Our model can forecast
individual customer''s future behavior accurately .(2) Our model
has the better ability to forecast more valuable customers than
the traditional regression model.
|