Collaborative Filtering Quiz

Practice Quiz: Collaborative Filtering

Question 1

Movie Rating Table:

Movie	Elissa	Zach	Barry	Terry
Football Forever	5	4	3	?
Pies, Pies, Pies	1	?	5	4
Linear Algebra Live	4	5	?	1

Assume numbering starts at 1 for this quiz, so the rating for Football Forever by Elissa is at (1,1)

What is the value of n_u?

Answer: 4 ✓

n_u represents the number of users. From the table, there are 4 users: Elissa, Zach, Barry, and Terry.

Note

Reference: From section 1 - “nu to denote the number of users” where in the example “nu is equal to four because you have four users.”

---

Question 2

What is the value of r(2,2)?

Answer: 0 ✓

r(i,j) = 1 if user j has rated movie i, and 0 otherwise. Movie 2 is “Pies, Pies, Pies” and user 2 is Zach. From the table, Zach has not rated “Pies, Pies, Pies” (shown as ?), so r(2,2) = 0.

Note

Reference: From section 1 - “r(i,j)=1, if user j has rated movie i” and “r(3,1)=0 because she has not rated movie number three.”

---

Question 3

In which of the following situations will a collaborative filtering system be the most appropriate learning algorithm (compared to linear or logistic regression)?

• You subscribe to an online video streaming service, and are not satisfied with their movie suggestions. You download all your viewing for the last 10 years and rate each item. You assign each item a genre. Using your ratings and genre assignment, you learn to predict how you will rate new movies based on the genre.

• You run an online bookstore and collect the ratings of many users. You want to use this to identify what books are “similar” to each other (i.e., if a user likes a certain book, what are other books that they might also like?) ✓

• You manage an online bookstore and you have the book ratings from many users. You want to learn to predict the expected sales volume (number of books sold) as a function of the average rating of a book.

• You’re an artist and hand-paint portraits for your clients. Each client gets a different portrait (of themselves) and gives you 1-5 star rating feedback, and each client purchases at most 1 portrait. You’d like to predict what rating your next customer will give you.

Collaborative filtering works best when you have multiple users rating multiple items, allowing the algorithm to find patterns across users and items. The bookstore scenario with many users rating many books fits this perfectly.

Note

Reference: From section 3 - “Because you have ratings from multiple users of the same item with the same movie. That’s what makes it possible to try to guess what are possible values for these features.”

---

Question 4

For recommender systems with binary labels y, which of these are reasonable ways for defining when y should be 1 for a given user j and item i? (Check all that apply.)

• y is 1 if user j has not yet been shown item i by the recommendation engine

• y is 1 if user j has been shown item i by the recommendation engine

• y is 1 if user j purchases item i (after being shown the item) ✓

• y is 1 if user j fav/likes/clicks on item i (after being shown the item) ✓

Binary labels should indicate positive engagement. Purchasing or liking/clicking after being shown an item represents user engagement and should be labeled as 1.

Note

Reference: From section 4 - “A labor of one means that the user engaged after being shown an item And engaged could mean that they clicked or spend 30 seconds or explicitly favorite or like to purchase the item.”