Thank you to Annelies Tjetjep, SAS Sr. Customer Account Executive, for contributing three tips to improve your use of SAS Enterprise Miner nodes and variables, as part of the SAS Enterprise Miner shortcut series.
Node tip #1: Did you change your mind about the role of a variable partway through a process flow? Use the Metadata node (found under Utility) to change only what you need wherever you want in the flow. Don’t worry, this won’t affect earlier nodes.
Node tip #2: Here’s a way to manipulate your variables to make them possibly perform better in the model: Use one of autonomous options in the Transform Variables node, like "Best," which tries several transformations for you then gives you the best one.
Node tip #3: If you have a variable whose behavior could easily change the outcome of the prediction (like age in a credit risk scenario) then you may not want to replace missing values with a constant number or an average that could easily miss vital information. In this case, you can use the Tree method in the Impute node which builds a decision tree model to predict missing values using all other non-missing input variables as inputs.
Have questions related to this tip? Ask them on the SAS Data Mining and Machine Learning Community to get perspective from a large pool of SAS Enterprise Miner experts. Simply click "New Message" (must be logged in!) and ask away.