You can set a filter in the analysis to make sure you are modeling with a subset of your data, and then apply your model on the records you set aside afterwards- the button for the filter feature is at the top right of the interface, once you’ve opened a dataset. Alternately, in the Model tab, you can change the Model Options in a toolbar to the far left of the interface, to use a sampling holdout. This will tell the software to use a random sample of the data for modeling, then validate against the remaining records automatically.
