Bagging and boosting both combine many decision trees, but they build them differently.
Which statement correctly contrasts the two?
Why did you pick that answer? Two or three sentences. The act of articulating it is what builds the judgment — not the click that follows.
A random forest (bagging) grows many deep, low-bias, high-variance trees on bootstrap samples and averages them, which mostly removes variance. Gradient boosting starts with a weak, high-bias model and adds small trees that fit the current residuals, steadily reducing bias — which is why boosting can overfit if you let it run and needs early stopping, while forests are hard to overfit by adding trees.
Boosting is the sequential one — each tree depends on the errors of the ensemble so far. Bagged trees never see each other.
Boosted trees are not independent; the whole point is that tree t is fit to the residuals of trees 1…t−1.
Averaging independent high-variance models reduces variance (bagging). Sequentially fitting residuals reduces bias (boosting).