160
Chapter 9
For more information, see the topic Connecting to the Repository on p. 161.
Storing and Deploying Repository Object s
Streams created in IBM® SPSS® Modeler can be stored in the repository just as they are, as files
with the extension .str. In this way, a single stream can be accessed by multiple u sers throughout
the enterprise. For more information, see the topic Storing Objects in the Repository on p. 164.
It is also possible to deploy a stream in the repository. A deploye d stream is stored as a file with
additional metadata. A deployed stream can take full a dvantage of the enterprise-level features
of IBM SPSS Collaboration and Deployment Services, such as automa ted scoring and model
refresh. For example, a model can be automatic ally updated at regularly-scheduled intervals
as new data becomes available. Alternatively, a set of streams can be deployed for Cham pion
Challenger analysis, in which streams are compared to determine which one contains the mos t
effective predictive model.
Youcan deploy a stream in one of two ways: as a stream (with the extension .str), or as a scenario
(with the extension .scn). Deployment as a stream enables the stream to be u sed by the thin-client
application IBM® SPSS® Modeler Advantage. Fo rm ore information, see the topic Opening a
Stream in IBM SPSS Modeler Advantage in Chapter 10 on p. 195. Deploy ment as a scenario
enables the stream to be used by Predictive Applications version 5, the predecessor of IBM
SPSS Modeler Advantage.
For more information, see Stream Deployment Options on p. 185.
Requirements for Streams Deployed as Scenarios
Toensure consistent a ccess to enterprise data, streams deployed as scenarios must be accessed
through the Enterprise View component of IBM® SPSS® Collaboration and Deployment
Services. This means that in SPSS Modeler, there must be at least one Enterpris eView source
node within each designated scoring or modeling branch in the stream.
Tou se the Enterprise View node, IBM SPSS Collaboration and Deployment Services
must be installed, configured and accessible from your site, with an Enterprise View,
Application Views, and Data Provider Definitions (DPDs) already defined. For
more information, contact your local administrator, or see the corporate website at
http://www.ibm.com/software/analytics/spss/products/deployment/cds/.
A DPD is defined against a particular ODBC data source. To use a DPD from SPSS Modeler,
you must have an ODBC data source defined on the SPSS Modeler server host that has the
same name and that connects to the same data store as the one referenced in the DPD.
In addition, the IBM® SPSS® Collaboration and Deployment Services Enterprise View
Driver must be installed on each computer used to modify or run the stream. For Windows,
simply install the driver on the computer where SPSS Modeler or SPSS Modeler Server is
installed, and no further configuration of the driver is needed. On UNIX ,a re ference to the
pev.shscript must b e added to the startup script. Contact your local adm inistrator for details
on installing the IBM SPSS Collaboration and Deployment Services Enterprise View Driver.