Practical Problem Solving - Java Data Mining: Strategy, Standard, and Practice - page 316

HEW Marketing
Manager

HEW
IT

TSB

Extract names and
addresses!

Build
Training
dataset and
train
clustering
models

2 weeks

Enhance
data

Customers

Compute
profiles on
enhanced data
and
buying data

Select one
clustering
Figure 12-3 Customer Segmenter scenario.

of the OMDB. This table could in turn be used by an online
analytical processing (OLAP) tool.
• HEW IT provides a table in the OMDB containing, for each
customer, the number of products bought per product line
in the past 12 months. In HEW, there are 10 product lines.
• HEW marketing manager then asks for profiles based on
the enhanced data as well, and on the buying behavior per
product line. For this, SQL count statements will be pro-
vided as reports to the marketing manager, who will
prepare a thorough report to the CMO.
• The CMO selects one clustering model and associates
cluster identifiers with names used for marketing purposes.

This process is presented in Figure 12-3.

12.3.2 Design of the CustomerSegmenter Object

Now that you've had success with the first two applications of JDM,
you are asked to design and implement the business layer for this
project. Some other team members will be in charge of the user inter-
face layer and the infrastructure layer. Of course, you will use the