Feature scaling in decision tree based models

Is feature scaling needed in decision tree based models? If no, I would quickly understand for classification, but what about decision tree regression models?

Hi @ECunha

I will share two angles on this:

  • Class imbalance
  • Feature scaling

Class imbalance
I’ve worked a few Decision Tree models where there is high class imbalance. I have to say that from my experience, Decision Trees are very good at handling class imbalance.

Feature scaling
As far as feature scaling, I usually apply feature scaling to all numeric features, regardless of the model, I think that it is a good practice to do so in every model because having features that are, for example, several magnitudes larger than others, or having ranges in a feature that are very wide, may affect the performance of the model.