For this project, our strategy for model estimation is to employ a complete SPSS Modeler stream developed from previous sections and then use SPSS Analytics Server for Spark Implementation. Our stream consists of SPSS Modeler Nodes for data treatment, as described in section, Data treatment and also model training Nodes with MLlib codes described in section, Methods for recommendation development as we prepared our SPSS Modeler to be ready to use MLlib in section, Spark for a recommendation engine.
As noted before, our IBM SPSS Modeler nodes created from Custom Dialog Builder depend on the Spark environment and will only run against IBM SPSS Analytic Server. SPSS Analytics Server is a tool to manage all the computing for model estimations and we have to employ IBM SPSS Analytic Server to implement the model estimation for this project, which makes everything easy for us. However, we also need to arrange for SPSS on the Spark system to run models for each movie category and also for each customer segment for us.
For more information about IBM SPSS Analytic Server, take a look at IBM Knowledge Center.
