Can someone help me check the answers to these data questions? 1 point In a regu

Question

Can someone help me check the answers to these data questions?

1 point In a regularized least squares problem, we have 10 features for each input x. Given the theta parameter estimations [0, 0, 2, 5, 0, 0, 0, 0,1, 0], what can we say about the model? * O It's more likely we used Ridge regression O It's more likely we used Lasso regression 1 point A model that achieves 0 training error will always outperform models with higher training error when evaluated on unseen test data.* O True False The cross-validation error is computed by averaging the error on all the training data and the test data. * O True O False 1 point Given a high dimensional n by d feature matrix, X, where d 1 point >n and with many zeros (e.g., constructed from bag-of- words features). Which of the following are true?* L1 Regularization can be used to determine informative dimensions We can still solve the normal equation: theta (XTX)(-1) X TY Adding L2 regularization will ensure that we can compute a solution to the least squares regression problem

Explanation / Answer

2. False - 0 training error means overfit model. On unseen test data, it might not be accurate. Models with low training error (non-zero) might be better.

Remaining are correct.

Thanks.

Related Questions

Navigate

• Browse (All)
• Browse C
• Subjects

---

---

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.