Chapter 10. Building the mining models using IM Modeling functions 227
Figure 10-1 Hybrid modeling run for classification of banking card holders
The ease of use with which you can apply a hybrid modeling process is covered
by the fact that data, models, and techniques are all available in a relational
database management system. Once you know how to use SQL and the SQL
API, you are basically up and running.
10.6 Conclusion
IM Modeling creates an infrastructure of tables, database objects, and stored
procedures to enable a table driven approach for creating data mining models.
This allows the mapping of a model produced by a data miner into SQL scripts.
These scripts can be automated and therefore ensure repeatable success once
the initial gems are found by the model.
Now that the model building process and the model are part of the database, and
because DB2 UDB is open, this data mining capability can be embedded in
applications using CLI, ODBC, JDBC, SQLJ, WebSphere MQ, and so on.
Analytical Data
Mart
Customers with
more than 86 debit
transactions,
remote access to
their account, and
have a least one
banking card
All clusters and
their attributes in
an overall view
Tree
induction
Demographical
clustering
Tree induction of
cluster with 26% of
banking card holders
Tree induction of a
cluster with 31% of
banking card holders
Customers with
less than 51 debit
transactions, only
teller access to
their account, and
have a least two
banking cards
228 Enhance Your Business Applications: Simple Integration of Advanced Data Mining Functions
Embedding data mining into applications also reduces the reliance on the data
mining expertise after the initial model building phase. Now there is a real
window of opportunity to embed advanced analytics on a wider scale than
previously.

Get Enhance Your Business Applications: Simple Integration of Advanced Data Mining Functions now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.