Error with Tree Ensamble Learner

KNIME Analytics Platform
1

Dear all,

I am experiencing an error with the Tree Ensamble Learner Node that I am unable to solve.

I have an unbalanced training set (category A 983, category B 5388), so I decided to modify the Ensamble Configuration settings so I can use equal size categories to derive each tree in the ensamble learner. The settings are as below:

image
image735×356 7.6 KB

Basically, for each tree I want to select all the samples from the minority category A and an equal number of random samples from the majority category B.

However, when I try to start the learner, I have the following error: “Execution failed: The parameter nrSelect must be > 0 but was 0”:

image

I am pretty sure that the error is caused by the “Equal size” sampling method, because when I set the “Data sampling Mode” to “Random” or “Stratified”, the node is executed correctly.

Do you have any suggestion how to solve this issue? I already used these settings in the past with data similar to those I am using right now, and I am not sure why it is not working this time…

Best

Domenico

2

Have you already tried it “with replacement” (boostrap)

3

This topic was automatically closed 182 days after the last reply. New replies are no longer allowed.