combination of decision trees?

SabaMomeniKho

Hello to all. Is there any way in rapidminer that we could combine multiple decision trees together in order to reach a more comprehensive tree (I guess) ?! I've heard that this cand be done for big data and I was wondering if it is possible to be done in rapidminer.
thank you 

varunm1

Hello @SabaMomeniKho

Does "group models" operator suffice your needs?

https://docs.rapidminer.com/latest/studio/operators/modeling/predictive/group_models.html

If you are looking for bagging and boosting there are "bagging" & "Adaboost" operators as well.

If you are looking for something more specific, please provide relevant details.

rfuentealba

Hello, @SabaMomeniKho

Another operator I normally use is "Vote" to mix a few decision trees and choose which one to use, besides AdaBoost, bagging and group models as my good friend @varunm1 already suggested.

It depends on what kind of interaction do you need on the models.

All the best,

Rod.

[Deleted User]

Hello

https://academy.rapidminer.com/learn/article/ensemble-models-diversity-works-like-magic

you can watch this video too

Regards,
mbs

[Deleted User]

@SabaMomeniKho

also you have this one

https://community.rapidminer.com/discussion/56823/adaboost-bagging-bayesian-boosting-classification-by-regression#latest

Good luck

BalazsBarany

Hi @varunm1,

be careful with Group Models. It's not applicable in this situation.
The help text says:

This operator groups the given models into a single combined model. When this combined model is applied, it is equivalent to applying the original models in their respective order.

Think about putting multiple decision trees into a process and connecting the *model* output of the first with the input of the second one. This will obviously fail.

Group Models is helpful for chaining preprocessing models (Normalize, PCA, Nominal to Numerical etc.) and then adding *one* predictive model. The result will be a combined model that does the preprocessing in the way it was set up, but can be handled (stored/retrieved/applied) as one model.

Regards,
Balázs

varunm1

@BalazsBarany oh yeah. Thanks for the correction. I thought more about bagging and boosting but some how suggested everything.

[Deleted User]

@SabaMomeniKho

If you need more information about this operators please let me know I will help you

Regards
mbs