How do I interpret the Random Forest Tree?

[LilianaGarcia]

Hi!

I’ve been trying to interpret the tree of the random forest predictor node but I don’t understand the origin of the branches. How do I interpret that? If someone know about a video or article that explain that, it will be great.

Thanks.

[ScottF]

Here are a couple of short videos from our KNIME TV channel on Youtube:

Random Forest

Random Forest Learner & Predictor

And a nice detailed blog post from our Evangelism team:

From a Single Decision Tree to a Random Forest | KNIME

Does this help?

[LilianaGarcia]

Thanks. It really helped me to understand more things.

[LilianaGarcia]

However, I want to understand precisely the tables that gave the random forest learner and predictor.

[ScottF]

Can you give me example of the table you need more help understanding? Sorry, it’s not clear to me. The attribute statistics, perhaps?

[LilianaGarcia]

Yes, sure. In this image is the example. It’s the random forest tree # 1 of the Titanic model.

In the Titanic model I tried to predict with the Randon Forest Learner and the Random Forest Predictor if a person survive or not. So when I see the view of the tree is confusing what is in bold. For example, the yellow box in the image show this "IsnotIn [Adults, Infant, Teenager, Senior], what is that mean? Because for all the attributes that I select in the node, is the same.

On the other hand, all the table for each node of the tree give the only results of 100% or 0% and that’s something that I truly do not understand, why is that?

[ScottF]

I think there may be a problem with the way you have set up the prediction workflow. Do you have an example workflow you can export and post here?

[LilianaGarcia]

This is the configuration of the Random Forest Learner:

And this is the set up of the Random Forest Predictor:

[mlauber71]

You set the minimum node size to 1 so it found a tree for -well- one observation.

You should prune your trees and limit the settings.