Fig. 12
Configurations of two regression-tree based ensemble learning models: bagging (a) and boosting (b). a In bagging methods, several regression trees are trained independently by their own subsets in which the data can be chosen more than once, and the final prediction is obtained by averaging the summation of all regression trees’ predictions. b The boosting method is generated sequentially, and each regression tree is related to the previous one. The final prediction is weighted median of all predictions of regression trees or summation of all predictions of regression trees.
