Diving Deep into Collaborative Filtering Techniques with ALS
This course explores collaborative filtering techniques, which are central to modern recommendation systems. It covers both user-based and item-based collaborative filtering methods, as well as matrix factorization and the powerful Alternating Least Squares algorithm.
Lessons and practices
Lesson 1: Exploring User-Item Explicit Rating Matrix
Loading Rating Matrix with NumPy
Adjust Missing Ratings Ratio
Handling Missing Ratings Randomly
Calculating Missing Data Proportions
Lesson 2: Implementing the Alternating Least Squares Algorithm
Initialize and Verify Factor Matrices
Update User Factors with ALS
Test and Evaluate Your ALS Predictions
Lesson 3: Understanding Implicit Feedback in Recommendation Systems
Building a Binary Interaction Matrix
Update Confidence with Logarithmic Scaling
Matrix Initialization from JSON Data
Normalize Watch Time for Certainty
Interpreting User Engagement Data
Lesson 4: Implementing Implicit Alternating Least Squares (IALS)
Create Preference and Confidence Matrices
Completing the Matrix Update Function
Top 5 Recommended Items
Lesson 5: Evaluating IALS Predictions Quality
Adjust Recommendations for Worse Metric
Create Normalized Item Rankings
Complete the Mean Rank Calculation
Evaluating Two User Recommendations
Interested in this course? Learn and practice with Cosmo!
Practice is how you turn knowledge into actual skills.