Feature importance after extrapolation- Why does it change

Intro to Machine Learning (2018)
1

Hello,

In lesson 5 under the topic of extrapolation, i observe that after extrapolation, i see that the importance of features changes. Why is that the case?

Earlier on, we observed that EROPS with AC or YearMade had a very high importance. Now with extrapolation, what resulted in the change in feature importances?

Any help would be very much appreciated.

Regards,
Kiran Hegde