SchoolWhool
Quizzes and saved progress, freeSign in freeSign in
Maths · Module 2Module material

Vectors interview questions

14 questions interviewers ask on this module’s topics, from first principles to the follow-ups. Each comes with a model answer.

  1. What is a vector in machine learning, and what is its dimension?
  2. What is the difference between shapes (d,), (1, d) and (d, 1) in NumPy?
  3. Name three norms and when you would use each.
  4. Why does lasso set weights to exactly zero and ridge does not?
  5. Why must features be scaled before k-nearest neighbours?
  6. What does the dot product tell you geometrically?
  7. When would you use cosine similarity instead of Euclidean distance?
  8. What is a projection, and where is it used?
  9. What does w · x + b = 0 describe, and what are w and b?
  10. Is a larger score w · x + b a larger distance from the boundary?
  11. A linear classifier gets 14 of 16 on a small dataset. Could another line do better?
  12. What is the curse of dimensionality?
  13. Does having 64 features mean nearest neighbours will fail?
  14. Why do decision trees struggle with diagonal boundaries?