Decision tree FAQs (Frequently Asked Questions)
My tree has tangled branches and/or overlapping nodes. How can I make it neater?
How can I add graphics to my decision tree model?
Can I create two or more trees in the same window?
How can a decision tree represent a fact like 'Weather IS Wet'?
Can a decision tree model represent uncertainty?
My tree has tangled branches and/or overlapping nodes. How can I make it neater?
Either use the 'Redraw neatly' command ('Build' menu) or else tidy the tree manually by dragging with the arrow or pen tools.
How can I add graphics to my decision tree model?
Use the 'Build' menu's 'Design' commands as explained here.
Can I create two or more trees in the same window?
Yes. Control-click with the pen to create a new root for each additional tree. See the 'Linked dtrees' example in the Examples Folder to see how two or more trees can work together to create a coherent model. The main idea is this: if at runtime InterModeller reaches a tree node labelled A which has an arc labelled V, and another tree exists in the same window which has a leaf label A=V, InterModeller automatically runs this other tree in order to select the appropriate arc of the first tree.
At runtime, InterModeller sometimes seems to offer more answer options than necessary when asking a question. Why?
Yes, this can happen. Suppose your tree has a question Q1 with three arcs containing answer options {A1, A2, A3}. You might expect that these should be the only options offered when Q1 is asked This is in fact the case if no other possible answers are ever associated with Q1. But suppose another node elsewhere on the tree contains the same question Q1, this time with arcs containing answers {A1, A4 and A5}. Then InterModeller will offer options {A1,A2,A3,A4,A5} whenever Q1 is asked. The advantage of this strategy is that the user gets a complete set (hopefully) of possible answers to the question to choose from, but it does mean that running a decision tree may fail to reach any conclusions.
How can a decision tree represent a fact like 'Weather IS Wet'?
Create a tree with only one node and label it 'Weather IS Wet'.
Can a decision tree model represent uncertainty?
Only rule models can incorporate certainty factors, so only they can express quantified uncertainty in the knowledge base. But at runtime, a decision tree model can certainly accommodate quantified uncertainty in the user data. Just make sure that certainty handling is enabled using the Run/Logic command before the model is run.