Boosting

In boosting, We create multiple different weak models. The next model is created on the basis of the previous model and especially to fix the error that was made by the previous model. And the final model's output is dependent on all of the previous model.

One of the most used Boosting model is Gradient Boosting.

The main difference with Bagging is that, in bagging, all the models are independent and here in boosting, models are created based on the error of the previous one.

• Boosting is more prone to Overfitting than Bagging
• Boosting is sequential, so cant be parallelized