07-11-2013 10:11 AM
I am using a data partition node inside a group node with a statification by segment. The Train and validation dataset are created as views. I would like to to create train and validation datasets for the various segments for further analysis purpose. How can i go about it.
I appreciate all your help.
07-12-2013 01:01 PM
When running group processing, you are right that any tables produced by the Partition node would get replaced with each loop. If you want to use each of the training and validation tables that are created during the loops, a SAS Code node is required. For example, you might have a flow such as IDS -> Start Group -> Partition -> SAS Code -> End Group, and the SAS Code node would contain code to create the training and validation data sets to use later.
PROC APPEND base = &em_lib.TrainbySegment data = &em_import_data force;
PROC APPEND base = &em_lib.ValidatebySegment data = &em_import_validate force;
This code would create two tables, TrainbySegment and ValidatebySegment. You could really put them anywhere you want.
I hope this helps. Another good resource for questions like this one is tech support: firstname.lastname@example.org.