The SPSS Model process
IBM® Campaign runs the SPSS Model process box as part of a flowchart. You can run the flowchart manually by using the flowchart controls, or you can automate the flow by using a Schedule process box in the flowchart. The model is built with current data that is extracted from the IBM® Campaign flowchart.
The SPSS Model process box must be properly configured and must point to a modeling stream in the IBM® SPSS® Collaboration and Deployment Services repository. Process boxes that supply input to the SPSS Model process box must have run successfully for data to be properly generated for a production run.
Deleting an SPSS Model process box does not delete any modeling streams it has created. IBM® Campaign deletes files that are created on the shared file system upon deletion of the SPSS® Model process box from the flowchart. To remove modeling streams, delete them in IBM® SPSS® Modeler Advantage Marketing Edition through IBM® SPSS® Collaboration and Deployment Services.
The following steps describe the SPSS Model process:
1.
When the SPSS Model process box is run, IBM® Campaign creates a data file by using the selected fields for the input cell.
2.
IBM® Campaign calls IBM® SPSS® Modeler Advantage Marketing Edition to build a model that uses the modeling stream in the IBM® SPSS® Collaboration and Deployment Services repository and the data file created in step 1. The input file overrides any data source that is configured in the modeling stream. If you build the model manually through the IBM® SPSS® Modeler Advantage Marketing Edition interface, whatever data source is configured in the modeling stream is used to build the model.
3.
IBM® SPSS® Modeler Advantage Marketing Edition builds the model and saves a file that is labeled LATEST in the IBM® SPSS® Collaboration and Deployment Services repository.
If the model is rebuilt because input data source variables were added or deleted, modeling configuration is preserved. If the configuration is valid, the model is built and saved in the repository as the latest version. If the configuration is not valid, an error is displayed, such as missing target variable.
4.
If IBM® SPSS® Modeler Advantage Marketing Edition returns a success message, the SPSS Model process box displays a green check.
If IBM® SPSS® Modeler Advantage Marketing Edition returns a failure message, the SPSS Model process box displays a red X.
5.
The data source is updated to point to the IBM® Campaign data that was used to build the model. The modeling stream in the repository is updated and contains a freshly built model.
Best practices for creating or rebuilding a model
Creating or rebuilding a model
Evaluating the model