Solving Operational Business Intelligence with InfoSphere Warehouse Advanced Edition

Chapter 10. Techniques for data mining in an operational warehouse 383

5. The scoring operation is what enables the developer to apply the mining

model to a data set for either batch or real-time operation. Almost all

operational BI scenarios use scoring.

Generally speaking, the data mining modeler uses the visualization and extractor

features to design and test the model. The scoring feature can be used to

validate the model against the test data set.

For an extensive discussion of data mining modeling in InfoSphere Warehouse

Design Studio, see InfoSphere Warehouse: A Robust Infrastructure for Business

Intelligence, SG24-7813.

10.2.4 Performing data mining scoring with other data mining models

through PMML

InfoSphere Warehouse 10.1 supports the execution (scoring) of data mining

models developed outside of InfoSphere Warehouse. These models can be

developed on either IBM or non IBM tools using a similar process flow as that of

InfoSphere Warehouse. The validated models are then exported to Predictive

Model Markup Language (PMML), a markup language similar to XML. The

PMML model can then be imported into InfoSphere Warehouse 10.1 and run on

DB2 data in batch or real time like any other data mining model.

The process of importing the PMML model is straightforward. Locate the target

database (the DB2 database on which you want the model to run) in the Source

Data Explorer of the InfoSphere Warehouse 10.1 Design Studio and expand the

Data Mining Models folder. Right-click the Data Mining Models folder and select

Import. The dialog prompts for the location of the PMML file and imports the

model into the DB2 database. After it is imported, it can be used like any data

mining model in DB2. For more information about using PMML models, see

InfoSphere Warehouse: A Robust Infrastructure for Business Intelligence,

SG24-7813.

10.3 Extended data mining techniques using SAS

Enterprise Miner

Thus far, we have described how data mining can be used in an operational data

warehousing environment. We have also described how the data mining

techniques in InfoSphere Warehouse 10.1 can be used to develop the mining

models that can be used in those scenarios. In this section, we consider

alternative modeling scenarios using tooling outside of InfoSphere Warehouse.

Specifically, we consider predictive data mining using SAS Enterprise Miner.

5