User-based vs item-based in ML Python - Practice Questions

Challenge - 5 Problems

🧠 Conceptual

intermediate

Difference in similarity calculation between user-based and item-based collaborative filtering

In collaborative filtering, similarity is a key concept. Which statement correctly describes how similarity is calculated differently in user-based versus item-based methods?

A User-based calculates similarity between items based on user ratings, while item-based calculates similarity between users based on item ratings.

B User-based calculates similarity between users based on their item ratings, while item-based calculates similarity between items based on user ratings.

C Both user-based and item-based calculate similarity only between users, ignoring items.

D Both user-based and item-based calculate similarity only between items, ignoring users.

💡 Hint

Think about who is being compared in each method: users or items.

User-based collaborative filtering finds users similar to the target user by comparing their item ratings. Item-based collaborative filtering finds items similar to the target item by comparing user ratings.

❓ Model Choice

intermediate

Choosing between user-based and item-based collaborative filtering for a large user base

You have a recommendation system with millions of users but only a few thousand items. Which collaborative filtering approach is generally more efficient and why?

A User-based, because it ignores item data and focuses only on user profiles.

B User-based, because millions of users provide more data to compute accurate user similarities quickly.

C Item-based, because it requires no similarity calculations.

D Item-based, because the number of items is smaller, so computing item similarities is faster and more stable.

💡 Hint

Consider which similarity matrix is smaller and easier to compute.

Item-based filtering computes similarity between items, which is smaller in number than users, making it more efficient and scalable for large user bases.

❓ Metrics

advanced

Evaluating recommendation accuracy for user-based vs item-based methods

You run both user-based and item-based collaborative filtering on the same dataset. Which metric would best help you compare their prediction accuracy on unseen user-item ratings?

A Total number of items recommended.

B Number of users in the dataset.

C Root Mean Squared Error (RMSE) between predicted and actual ratings.

D Training time of the model.

💡 Hint

Think about a metric that measures how close predictions are to real ratings.

RMSE measures the average difference between predicted and actual ratings, making it suitable to compare accuracy of recommendation models.

🔧 Debug

advanced

Identifying the cause of sparse similarity matrix in user-based filtering

You implemented user-based collaborative filtering but notice the user similarity matrix is very sparse, causing poor recommendations. What is the most likely cause?

A Users have rated very few common items, so similarity cannot be reliably computed.

B The model is overfitting due to too many training epochs.

C The dataset contains too many items, causing memory overflow.

D The item similarity matrix was used instead of the user similarity matrix by mistake.

💡 Hint

Think about what is needed to calculate similarity between two users.

User similarity requires overlapping rated items. If users have few or no common rated items, similarity scores are sparse or zero.

🧠 Conceptual

expert

Impact of cold start problem on user-based vs item-based collaborative filtering

Which statement best explains how the cold start problem affects user-based and item-based collaborative filtering differently?

A User-based struggles more with new users who have no ratings, while item-based struggles more with new items that have no user ratings.

B User-based struggles more with new items, while item-based struggles more with new users.

C Both user-based and item-based methods are equally unaffected by cold start problems.

D Cold start only affects content-based filtering, not collaborative filtering.

💡 Hint

Consider what data each method needs to make recommendations.

User-based filtering needs user rating history, so new users with no ratings cause problems. Item-based filtering needs item rating history, so new items with no ratings cause problems.