topic Re: partitioning data by a nominal variable in SAS Data Science

partitioning data by a nominal variable

DocMartin — Tue, 26 Jul 2016 14:54:10 GMT

The partition mode in EM lets one split up a data set into training, validation, and test data sets base on random sampling. However, I want to partition my data by the values in a variable I call datset: 'D', 'V', 'T'. How can I achieve this using EM? Thanks! Andrew

Re: partitioning data by a nominal variable

WendyCzika — Tue, 26 Jul 2016 17:11:01 GMT

You can do it with a SAS Code node with the following code:

data &EM_EXPORT_TRAIN &EM_EXPORT_VALIDATE &EM_EXPORT_TEST;
   set /* your data */;
   if datset='D' then output &em_export_train;
   else if datset='V' then output &em_export_validate;
   else if datset='T' then output &em_export_test;
   drop datset;
run;