I don’t believe this has been covered earlier; the parameters used in the solution appear to be different from those that appeared in earlier notebooks. So how are we to make the determination of what the most important parameters are? And, could these not, in fact, be different for different problems?
If the bagging is used with decision tree, then one will usually use a RandomForest instead and this topic is later covered: Hyperparameter tuning — Scikit-learn course
If bagging is used with another base estimator, then the important hyperparameters will be linked to this specific base estimator.
The bagging itself has only the number of estimators as an important parameter.
1 Like