Table 2.

Summarized differences between Bagging and Boosting.

Bagging	Boosting
Weak models often learn independently in parallel	Weak models often learn sequentially in an adaptive way
Bagging focuses on obtaining an ensemble model with less variance	Boosting focus on producing a strong model with less bias but variance can also be reduced
Different weak learners can be fitted independently and train concurrently	Different weak learners cannot be fitted independently but models are fitted iteratively and training of each model depends on the model fitted previously
The idea behind boosting is to construct a set predicting model by aggregating the results of base models.	The idea behind boosting is to construct a set of models which are aggregated to get a strong learner.