#boosting

Question 1: Single choice

Which statement about random forests and averaging is the most accurate?

Each tree first draws a random subset of features and then uses only those features for all of its splits1
The number of trees should be tuned like other hyperparameters to get better validation scores2
Random forests build their trees one after another, each depending on the previous tree3
Every tree in a random forest is trained on a different bootstrap sample of the training set4
Averaging classifiers with training errors of 10%, 20% and 30% gives at best a training error of 10%5