The School of Computer Science Presents....
Recommendation Systems in Python (Part 2)
Presenter: Shaghayegh Sadeghi
Date/Time: Wednesday, October 18th, 4:00pm – 5:00 pm
Location: 4th Floor (Workshop space) at 300 Ouellette Avenue (School of Computer Science Advanced Computing Hub)
LATECOMERS WILL NOT BE ADMITTED once the presentation has begun.
In this workshop, students will learn everything they need to know to create their own recommendation engine. Through hands-on exercises, students will get to grips with the two most common systems, collaborative filtering, and content-based filtering. Next, students will learn how to measure similarities like the Jaccard distance and cosine similarity, and how to evaluate the quality of recommendations on test data using the root mean square error (RMSE).
By the end of this course, students will have built their very own movie recommendation engine and be able to apply their Python skills to create these systems for any industry.
Collaborative Filtering
Collaborative filtering
• Pivoting your data
• Finding similar users
• Challenges with missing values
• Compensating for incomplete data Finding similarities
• User-based to item-based
• Similar and different movie ratings
• Finding similarly liked movies
Using K-nearest neighbors
• Stepping through K-nearest neighbors
• Getting KNN data in shape
• KNN predictions
Item-based or user-based
• Comparing item-based and user-based models
• Which one to choose?
Matrix Factorization and Validating Your Predictions
Dealing with sparsity
• Matrix sparsity
• Limited data in your rows
• Matrix multiplication
Matrix factorization
• Identifying latent features
• Information loss in factorization
Singular value decomposition (SVD)
• Normalize your data
• Decomposing your matrix
• Recalculating the matrix
• Making recommendations with SVD
Validating your predictions
• Calculating RMSE
• Comparing recommendation methods
Knowledge of Supervised Learning with scikit-learn and pandas
Shaghayegh is a Ph.D. student at the School of Computer Science (University of Windsor). Her main research interest is in biological network embedding.