Movie Recommender System — ALS, Collaborative Filtering & Market Segmentation

Link to the complete documentation.

This project was built and run entirely on Databricks (PySpark), not locally — so there's no requirements.txt and no scripts to run in sequence. The notebook was exported as .ipynb and lives in src/ as a record of the full pipeline; it won't execute standalone outside a Spark/Databricks environment.

Structure

• src/ — the exported Databricks notebook(s), covering the full pipeline: data loading & EDA, train/val/test split, popularity baselines, item- and user-based collaborative filtering, ALS (explicit and implicit), and Jaccard/MinHash market segmentation
• docs/ — write-up of the methodology, math derivations, and results for each section

To Reproduce

1. Create a Databricks workspace (Free Edition is sufficient)
2. Download MovieLens ml-latest-small and upload ratings.csv, movies.csv, tags.csv to a Unity Catalog Volume
3. Import the notebook(s) from src/ into your workspace and run cells top to bottom — each section depends on DataFrames/models built in earlier cells within the same notebook