#n-estimators

Question 1: Single choice

mediumAI-checkedDecision Trees & Ensembles

A gradient-boosting model reaches a good validation score with learning rate $0.1$ and 200 trees. You lower the learning rate to $0.05$ . What should you expect?

You will need half as many trees, since each tree is now more accurate1
You will need roughly twice as many trees to reach a similar training fit2
The trees will be grown deeper automatically to compensate3
The number of trees no longer matters, because a small learning rate prevents overfitting4

Timed test