#extra-trees

Question 1: Single choice

hardAI-checkedDecision Trees & Ensembles

Instead of searching every threshold of each candidate feature for the best split, a tree ensemble draws one random threshold per candidate feature and keeps the best of these random splits. What is this method, and what is its usual effect compared with a random forest?

Bagging; each tree sees a bootstrap sample but uses exhaustive splits1
Extremely randomized trees; training is faster and variance drops a bit more, at a small cost in bias2
Extremely randomized trees; they always overfit more than a random forest because every split is random3
Gradient boosting; each tree corrects the residual errors of the previous ones4

Timed test