It's straightforward with Enterprise Miner and you can try this out by using the Fisher-Anderson Iris data, which is provided as a sample dataset:
1. In Help > Generate Sample Data Sources, choose the iris data and press OK.
2. Drag the Iris data onto the canvas.
3 Make sure Species is defined as a nominal target. (Confirm by right clicking on the node and choose Edit Variables...)
4. Connect it to a Decision Tree node and run. You will get a tree whose terminal nodes essentially correspond to different species (setosa, virginica, versicolor).
In actual practice you would normally partition the data into training and validation samples before running the tree. You can use the Data Partition node for that.
You can also look at the HP Tree node for distributed processing.
Hope this helps.
(note: edited for clarity)
