Setting Random Forest node parameters

KNIME Analytics Platform
1

Hey there,

i want to classify data by random forest. The implementation works. But i have a question: A random forest is built by two parameters, ntree trees and mtry variables. Where can i configure the mtry variables.
The number of ntree trees can be configured with the option ‘number of models’. But which option or configuration is equal to mtry variables. Where can i set this parameter. I thought about ‘limit number of levels’ and ‘minimum node size’ but i don’t think that this is what i am looking for.

Thank you for your help
Greetz
Nik

2

Hi Nik

the ntree is the number of models used for learning the tree? Than you can find this at the button of the dialog under Forest Options and Number of Models.

The mtry is the number of variables available for selection for each of the splits. This cannot be changed in the Random forest Learner node, it will always take sqrt(m) number of features, where m is the number of attributes.

If you need more options, this number you can edit in the Tree Ensemble Learner node.

Cheers, Iris

2 Likes