This works better for the classification tasks. From all the prediction of the weak learners, we take the one with the majority votes and in case tie, randomly pick one.

The classification counterpart of Averaging in Ensemble Learning; used in Bagging and Random Forest.