Partition Question

[AnnaGeorge1]

How do you use a user-defined partition in Enterprise Miner?

[gcjfernandez]

You can use data partition node and input user-defined train, validation and test percentages:

[AnnaGeorge1]

I am not asking how to partition the data. My data set has a validation column where the observations are already identified as test, train, or valid. What I want to know is how I can use that to communicate with Enterprise Miner that the data are already partitioned.

[gcjfernandez · Accepted solution]

First separate training and validation data based on partition indicator variable in SAS EG or using SAS code node and save to project Library.

Then using SAMPLE -> Input data node add these two data into the process flow, and assign their respective roles (Train or valid).

[MelodieRush]

You can use the following code in the SAS Code node to divide your data into training, validation and test datasets to be used for Enterprise Miner nodes

data &EM_EXPORT_TRAIN &EM_EXPORT_VALIDATE &EM_EXPORT_TEST;

set &EM_IMPORT_DATA; 

if partition_key=1 then output &EM_EXPORT_TRAIN;

else if partition_key=2 then output &EM_EXPORT_VALIDATE;

else if partition_key=3 then output &EM_EXPORT_TEST;

run;

Put this code in the training code section which is the top icon selected in the image below.

Your flow would/could look something like this: