02-08-2012 01:01 PM
I'm using EM 7.1 and the variable clustering node gives an error if the dataset has over 100,000 observations. I can sample the dataset, and then cluster my variables, but is there an easy way to pass along only the new clustered variables along with the full dataset (train, validate, test) to the models?
02-22-2012 05:34 PM
On the options for the node, look under "Stopping Criteria". There is an option called "Suppress Sampling Warning". Set it to "Yes" and your problem should be solved.