2020W2 midterm solutions PDF

Preview

Summary

2020 midterm solutions for parctice and review...

Description

2/27/2021

Midterm

 Students have either already taken or started taking this quiz, so be careful about editing it. If you change any quiz questions in a significant way, you may want to consider regrading students who took the old version of the quiz.

Points 22  Published

Details



Questions

Show Question Details

 MCQ

Pick 13 questions, 1 pts per question

 Question What does xij refer to in the notation of the course?

t Answer

 i-th sample of the j-th feature in the training data  i-th feature of the j-th sample in the training data  i-th sample of the j-th feature in the test data  i-th feature of the j-th sample in the test data

 Question Consider fitting an ordinary least squares linear regression model (without an intercept) on a dataset. After training, you end up with the following weight vector: w = [-1] [ 3] [ 2]

From this w, what can you conclude about the number of training examplesn and the number of featuresd?

https://canvas.ubc.ca/courses/58981/quizzes/324855/edit

1/10

2/27/2021

Midterm

 n=1; d=3  n=3; d=1  not enough information to determine n; d=1 t Answer

 not enough information to determine n; d=3  n=1; not enough information to determine d  n=3; not enough information to determine d

 Question What is the time complexity of assigning a single test example to a cluster using k-means clustering?

 O(ndk)  O(nd) t Answer

 O(dk)  O(n^2d)  O(n^2dk)  O(nk)

 Question Which of the following models would be the most reasonable case to use gradient descent for training?

 DBSCAN  k-means clustering  KNN t Answer

 polynomial regression with smoothed absolute value loss

https://canvas.ubc.ca/courses/58981/quizzes/324855/edit

2/10

2/27/2021

Midterm

 least squares linear regression with RBF features  random forest  naive Bayes  linear regression minimizing the L1 loss, ||Xw-y||_1...