KNN Recommender

By using MovieLens Datasets we build a recommender system based on KNN Item-Based Collaborative Filtering.

Enviroment Setup

Use poetry to set up the Python environment (Python >= 3.9)

Get the full data set from MovieLens Datasets and put the movies.csv and ratings.csv in this repo.

Complete the exercises

Find all the TODO comments in knn_recommender.py and finish the missing pieces in the code. The finished version is in the finished branch

Run in CLI

Run the knn_recommender.py with the following optional options:

• --movie_name : provide your favoriate movie name
• --top_n : top n movie recommendations
• --path : input data path, default to be ./
• --movies_filename : default to be movies.csv
• --ratings_filename : default to be ratings.csv

Example:

python knn_recommender.py --movie_name "Iron Man" --top_n 10

Run in Browser

Make sure you have run it once in the CLI and have the pickle files (hashmap.p and movie_user_mat_sparse.p) in this repo

Start a local webserver:

python -m http.server

Open http://0.0.0.0:8000 and select knn_recommender.html from there

Presentation slides

Visit the slides on slides.com

---

Credit to Kevin Laio (KevinLiao159) for original code and blog post