Gradient
Lesson 3.7ProjectCode~2 h

Project: a movie recommender

Users and movies as vectors: find similar people with cosine similarity and predict what someone will like.

Module 3 · 0/7 done

Project: a movie recommender

Every streaming service has to guess what you’ll like next. In this project you’ll build one from vectors, dot products and PCA. The data: 20 people, 10 movies, and the ratings (1–5 stars) they gave. About a third of the table is blank: movies people haven’t seen yet.

  1. Measure how similar two people’s tastes are.
  2. Predict missing ratings from similar people, and prove it beats a simple baseline.
  3. Draw a map of the movies, using nothing but who liked what.