Extra Trees

Hello everyone,

Is it possible to use the algorithm Extra Trees (Extremely Randomized Trees) with KNIME?
I have tried to change the settings of the node “Tree Ensemble Learner” to use it as Extra Trees, but i don’t know if this is a possible way.

Thanks in advance,

Hi @may_paludetti,

interesting question :slight_smile:

According to the following article Extra Trees use a randomized node split, which is not possible with the “Tree Ensemble Learner” https://www.baeldung.com/cs/random-forest-vs-extremely-randomized-trees#:~:text=Extremely%20Randomized%20Trees%2C%20also%20known,with%20different%20subsets%20of%20features

In case you can’t reach a satisfactory model with the Tree Ensemble Learner node or the Random Forest Learner node and you would like to try the Extra Trees model, you can use a Python Snippet node with the ExtraTreesClassifier function from sckit-learn:



@may_paludetti as it happens I was toying around with some workflows and also ChatGPT so I put it to the test for this one and adapted another workflow to handle the suggested sklearn with extra-trees classifier. Since that would only like numeric variables I have employed the vtreat package to handle that quickly. You can use your own data preparation and settings of course.