yilinee / science-of-netflix

Slides, collaboration doc, and code building a recommendation system example presented in a meetup talk delivered April 22nd, 2020

The Science of Netflix

A code along workshop for the Flatiron School Meetup The Science Behind Netflix Recommendations : Talk | Chicago. The original materials for this workshop were developed by Flatiron NYC Coach Yish Lim and were adapted for online Zoom broadcast by David John Baker and Ben Oren.

This version of the workshop is being presented on:

• April 22nd, 2020

The slides from the presentation are available here and the collaborative document in use during the broadcast is here. There will be a link here to the video recording made of the talk as well.

Here is a binder you can launch the notebook from.

The dataset used is The Movies Dataset found on Kaggle, specifically from the file ratings small.csv: https://www.kaggle.com/rounakbanik/the-movies-dataset

If you do want to run the code on your own, download Jupyter Notebook and run these lines in your terminal:

1. git clone https://github.com/ben-oren/science-of-netflix.git
2. cd science-of-netflix
3. jupyter notebook

Additional Readings and Resources:

Yish has written a couple posts on Medium about recommendation engines on her blog:

• A Primer to Recommendation Engines
• Collaborative Filtering on Utility Matrices

There are several freely available textbooks which mention recommendation systems. The first link is a good general introduction; the second one goes into a deeper dive of the math underlying SVD using gradient descent + alternating least squares as demonstrated in the talk.

• https://link.springer.com/book/10.1007%2F978-3-319-29659-3
• http://infolab.stanford.edu/~ullman/mmds/book.pdf

Here are some other links to podcast episodes on recommendation engines:

• Data Skeptic: http://dataskeptic.libsyn.com/recommender-systems-live-from-farcon-2017
• Linear Digressions: http://lineardigressions.com/episodes/2017/6/25/factorization-machines

Feel free to reach out to Yish, Dave or Ben!