Question 2 should be reworded.
This question is extremely simple and yet I got the wrong answer.
Because Decision trees do a constant extrapolation, they perform extremely badly. On that we can all agree.
Still, should we consider that they are only capable of interpolating ?
We can rework this part in the next version of the MOOC. The question/answer is still in line with the conclusion of the previous exercise M05.2 but it is indeed not entirely correct.
Hi, I add a +1 to the remark of Quentin; I also took the answer a as the TreeRegressor makes it possible to predict constant values outer of the dataset range (even if it is a dummy strategy)