Question 3 Quiz

Module 6. Ensemble of models Ensemble method using bootstrapping
#1

The answer says a) and c) because “A random forest is a bagging using trees. It includes an additional bootstrap on the features”.
I’m confused by this answer because:

  • From the notebooks I understood that Random Forest Classifier includes an additional bootstrap on the features but no Random Forest Regressor. Thus c) wouldn’t always be right.

  • In the documentation for sklearn.ensemble.BaggingClassifier and sklearn.ensemble.BaggingRegressor there is the parameter bootstrap_features so you can choose wether or not to do this. In this case also c) wouldn’t be right.

Also the answer is somewhat ambiguous because it doesn’t precise if a), b) and c) are characteristics of bagging predictors or random forests.

Hope it helps!

#2

This is tracked in github.

#3

Thanks very much @Consuelola for the very valuable feedback. Indeed we discovered several errors and sources of confusion in the related notebooks and we decided to rework them. You can have a look a the new version here:

https://inria.github.io/scikit-learn-mooc/python_scripts/ensemble_bagging.html
https://inria.github.io/scikit-learn-mooc/python_scripts/ensemble_random_forest.html

You should be able to synchronize the notebooks on the fun platform using the “File” menu of Jupyter if you wish.

The errors in this question of the quiz has also been udpated based on your feedback to remove any source of ambiguity. The quiz will be synchronized soon.

#4