Pseudo Random Numbers Generator

Dear community,

I have inquired on the matter which random numbers generator(s) KNIME is using in general,

and in particular in the bagging process of the RandomForestLearner Node. Is it by any chance


I intend to reproduce a finding in sci-kit learn in KNIME and need exact information about the implementation/

default settings of the node.


Furthermore, will the GINI impurity splitting criterion be implemented any time soon?

Last but not least, is there a more thorough documentation of the RandomForestLearner node?


Kind regards,


Hello moxplayer,

I believe most nodes use java.util.Random as random number generator, at least the random forest does.

Exactly reproducing results of a different implementation could proof to be very difficult as there are many factors at play.

Regarding your second question: The gini impurity is already implemented in the KNIME random forest (and tree ensemble) nodes for classification. You have to select Gini Index as splitting criterion. I agree that the naming is somewhat wrong here and I will file a bug to change this.

Kind regards,