#collinearity

Question 1: Single choice

hardAI-checkedLinear Models & Regularization

A dataset contains three identical copies of one feature, and the best fit needs a total weight of 1 on that feature. How do ridge (L2) and lasso (L1) regression share this weight among the three copies?

Ridge splits it equally, 1/3 each; lasso's split is not unique1
Both split it equally, 1/3 each, because the copies are identical2
Ridge puts all weight on one copy; lasso splits it equally3
Neither model can be fit, since collinear features make the solution undefined4

Timed test