Question

Which among the following prevents overfitting when we perform bagging?
A) The use of sampling with replacement as the sampling technique
B) The use of weak classifiers
C) The use of classification algorithms which are not prone to overfitting
D) The practice of validation performed on every classifier trained

Answer 1

The correct answer is B that is the use of weak classifiers which prevents overfitting when we perform bagging. In bagging, the outputs of multiple classifiers trained on different samples of the training data are combined which helps in reducing overall variance. Normally unstable classifiers can be made robust because of reduction in variance that is bagging reduces variance of the classifier.

Answer 2

The option B is the correct answer.

The use of "weak classifiers" prevents over fitting when we perform "bagging". When over-training is present that causes the over fitting, will not generate any issue with weak classifiers.

For instance, when a decision tree has one node, consider that as the "root node", there is no scope for over fitting. In turn, it assists the classifier to combine with the weak classifiers outputs to "avoid over fitting."