Module 5 Assessment: Support Vector Machines Check

Estimated active time: 30-45 minutes

Question 1

How much does the score move across your C and gamma grid?

Pass answer: reports the spread, not only the best cell, and treats sensitivity as the finding.

Question 2

Why must features be scaled before fitting an SVM?

Pass answer: explains that the margin is defined by distance, so the widest-range column dominates.

Question 3

What did the RBF kernel buy over the linear one here?

Pass answer: compares both figures and says whether the gain justifies the tuning cost.

Question 4

What would you report if the best grid cell were on the edge of your search?

Pass answer: recognises the grid should be extended rather than the edge value accepted.