Tree Ensemble Tree View statistics

Dear KNIME-Team,

I'm having an issue with the statistics in the tree ensemble learners tree view. It always shows n = 1 and always 100% for one of the 2 classes. It's hard to explain so please look at the attached screenshot and workflow to create it. For me it looks like a bug or what am I missing?

I would expect to see the class distribution before each split and afterwards.

 

 

 

Hello beginner_,

there is an option in the dialog to enable the class distributions in the view:

Tree Ensemble Learner: Attribute Selection -> Misc Options -> Save target distribution in tree nodes

Random Forest Learner: Options -> Misc Options -> Save target distribution in tree nodes

The reason why we don't store this by default is that it can have a considerable impact on the model size (in terms of memory) and is not needed for simple prediction.

Please note that decision tree views in random forests are different to normal decision trees because they are trained on a bootstrap sample of the data and at each node only a random subset of the features is evaluated for possible splits.

 

Cheers,

nemad