The SPSS® Score process
Unica Campaign runs the SPSS® Score process box as part of a flowchart. The individuals to be scored and their data come from the source cell that is defined in the SPSS® Score process box.
The SPSS® Score process box must be properly configured and point to at least one modeling stream in the IBM SPSS Collaboration and Deployment Services repository.
The SPSS® Score process can score using up to five different models run against the same input cell. You might use multiple models to compare response likelihood against different offers, one model for each offer. For example, score one model to generate the likelihood of response and score another model to predict the magnitude of the purchase amount.
If you delete an SPSS® Score process box, Unica Campaign cleans up any associated files from the shared folder and from the partition/<partition n>/tmp folder.
File length errors
If you frequently see
the message Delimited file length exceeds allowable maximum
length
when you run the SPSS® Score process box, you can increase the
value of the configuration parameter.
- In Unica Campaign configuration, navigate to Affinium|Campaign|partitions|partition1|server|internal.
- Change numRowsReadToParseDelimitedFile to
a number higher than the default value of 100.
The value depends on the total number of records to be processed in the SPSS® Score process box. For any value for this configuration property, half of the records are processed from the top of the score output file, which is received from IBM SPSS Modeler Advantage Enterprise Marketing Management Edition. The remaining half of the records are processed from the bottom of the file to determine maximum field width for all fields.
If you set this configuration parameter to a high value, you can process more records to help ensure accurate identification of the maximum field width for all fields.
If you set this configuration parameter to 0, the entire file is scanned to accurately determine the field width of all fields. This action resolves the error but increases the SPSS® Score process run time if the file size is large.
The following steps describe the SPSS® Score process:
- When the SPSS® Score
process is run, Unica Campaign specifies
the following information:
- The modeling stream in the IBM SPSS Collaboration and Deployment Services repository
- The input and output file paths for the data to be scored
- The scoring results, such as from a Snapshot process box
- Unica Campaign creates an input file for the audience members to be scored. The file contains the fields or variables that are required by the models.
- The SPSS® Score process runs. The results are stored in a file in a shared directory on the server. Unica Campaign automatically retrieves and makes scored fields available to downstream process boxes as persistent derived fields.
- The downstream processes can use the output scores to further
subselect or segment the data or to take a snapshot of the score values. Note: If the score outputs that are used by the downstream processes are not available from the SPSS® Score process, the downstream processes fail. This failure might happen, for example, if the score configuration was changed in the modeling stream and different outputs were defined.