The correct answer is: A. Boosted trees
Boosted trees are a type of machine learning model that uses a series of weak learners to produce a strong learner. The weak learners are trained on different subsets of the data, and their predictions are then combined to produce a final prediction. The importance of each weak learner is calculated, and the most important learners are given more weight in the final prediction.
Bagged trees are a type of machine learning model that uses a series of decision trees to produce a single prediction. The decision trees are trained on different subsets of the data, and their predictions are then averaged to produce a final prediction. The importance of each decision tree is not calculated.
Partial least squares is a statistical method that is used to find linear combinations of variables that are most correlated with a response variable. The importance of each variable is calculated, but it is not used to produce a final prediction.
Therefore, the only model that sums the importance over each boosting iteration is boosted trees.