Boosting is a machine learning ensemble method for improving the accuracy of machine learning algorithms by reducing bias. It is used for combining the predictions of various single algorithms. Boosting is a sequential method that iteratively adjusts the weights of observations.
Boosting differs from bagging in a way that it uses the previous model in the process, while models are built independently in bagging. Also, boosting is aimed to decrease bias instead of variance, as it happens in bagging. Finally, bagging assigns equal weights to models; boosting, on the other hand, assigns weights based on the model’s performance.