Decision Tree: Pre-pruning and Pruning settings

pennypolarbear

Trying to do a churn analysis. Following the parameters in the example process, only one node appears in the tree.

Are there any ways to optimally adjust the parameters for pre-pruning (minimal gain etc.) to enlarge the tree as well as prevent over-fitting?

 

IngoRM

Hi,

 

Sometimes it is hard for a decision tree to separate the classes and the tree collapses into a single node.  Please check the following article for ideas how to change parameters to avoid this.  Alternatively, you can also try a different learner of course.

 

http://community.rapidminer.com/t5/RapidMiner-Studio-Knowledge-Base/My-Decision-Tree-Shows-only-one-...

 

Hope this helps,

Ingo