Collaborative filtering in ML Python - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Understanding User-Based Collaborative Filtering

In user-based collaborative filtering, how are recommendations typically generated?

A By analyzing item features and recommending items with similar attributes

B By clustering items and recommending items from the largest cluster

C By randomly selecting popular items from the dataset

D By finding users similar to the target user and recommending items they liked

💡 Hint

Think about how users' preferences help suggest new items.

User-based collaborative filtering finds users with similar tastes and recommends items those similar users liked but the target user hasn't seen yet.

❓ Predict Output

intermediate

Output of Cosine Similarity Calculation

What is the output of the following Python code that calculates cosine similarity between two user rating vectors?

ML Python

import numpy as np
from numpy.linalg import norm

user1 = np.array([4, 0, 3, 5])
user2 = np.array([5, 1, 2, 5])
cos_sim = np.dot(user1, user2) / (norm(user1) * norm(user2))
print(round(cos_sim, 2))

A 0.85

B 1.00

C 0.97

D 0.75

💡 Hint

Calculate dot product and norms carefully.

The cosine similarity is the dot product of the vectors divided by the product of their magnitudes. The result rounds to 0.97.

❓ Model Choice

advanced

Choosing the Best Collaborative Filtering Model for Sparse Data

You have a very sparse user-item rating matrix with many missing values. Which collaborative filtering approach is generally better suited to handle this sparsity?

A Item-based collaborative filtering with cosine similarity

B Matrix factorization techniques like Singular Value Decomposition (SVD)

C Random recommendation based on item popularity

D User-based collaborative filtering using simple nearest neighbors

💡 Hint

Consider methods that can learn latent features from sparse data.

Matrix factorization methods like SVD can learn latent factors and handle sparsity better than neighborhood methods, which rely on overlapping ratings.

❓ Hyperparameter

advanced

Effect of Number of Latent Factors in Matrix Factorization

In matrix factorization for collaborative filtering, what is the effect of increasing the number of latent factors (dimensions)?

A It can improve accuracy but may cause overfitting if too large

B It always improves model accuracy without any downside

C It reduces model complexity and speeds up training

D It has no effect on the model's performance

💡 Hint

Think about model complexity and generalization.

Increasing latent factors allows the model to capture more complex patterns but risks overfitting if the number is too high relative to data size.

❓ Metrics

expert

Evaluating Collaborative Filtering with RMSE

You trained a collaborative filtering model and computed the Root Mean Squared Error (RMSE) on a test set. Which statement about RMSE is correct?

A A lower RMSE means the model's predictions are closer to actual ratings

B RMSE measures how many recommendations are correct

C A higher RMSE indicates better model performance

D RMSE is not suitable for rating prediction tasks

💡 Hint

Recall what RMSE measures in prediction tasks.

RMSE measures the average magnitude of prediction errors; lower values mean predictions are closer to true ratings.