a hands-on course

Build a Movie
Recommender

(Collaborative Filtering)

Build the 'people like you also watched' engine behind Netflix — in NumPy, in about half an hour.

A free, hands-on tutorial: build a movie recommender from scratch in NumPy using collaborative filtering — the 'people like you also watched' algorithm behind Netflix and Amazon. Find similar users with cosine similarity, predict ratings, and rank recommendations. Runs entirely on your laptop, no API key.

in the real world

Collaborative filtering is the classic engine behind Netflix, Amazon and Spotify recommendations — the same 'people like you' idea that decides what millions of feeds show next.

Start with the ratings →

Part 1 — Taste as data

1

The ratings matrix

Install NumPy and build the users-by-movies grid of ratings — with all the blanks a recommender exists to fill.

2

Finding similar tastes

Measure how alike two people's taste is with cosine similarity — after subtracting each person's rating bias.

Part 2 — Make recommendations

3

Predicting a rating

Guess how a user would rate a movie they haven't seen — a similarity-weighted average of what similar users thought.

4

Recommending movies

Predict every unseen movie for a user, rank them, and return the top picks — a real recommendation.

5

Because you liked…

Flip the idea: find movies similar to each other, so you can explain a rec with 'because you liked The Matrix'.

Part 3 — Judge it and go further

6

Evaluate & what's next

Hide a rating you know, predict it, and see how close you got — plus cold start and where recommenders go next.