Which of the following statements does NOT describe a characteristic of bagging?

Group of answer choices

Bagging provides predictions based on the majority vote of a committee of base learners whose predictions are no better than random guesses.

Bagging includes built$-$in out$-$of$-$sample error estimation so additional cross$-$validation is unnecessary.

Bagging generates a committee of different learners by constructing them from slightly different training sets.

A weakness of bagging is that the individual models still may be highly correlated.