MovieLens Recommender System Project

Overview

This project was completed for the course Analysis of Customer Data.
The objective was to analyze customer rating data from the MovieLens dataset and build recommender system models to compare against a baseline.

The project includes two key components:

1. Data exploration and descriptive analysis
2. Development, tuning, and evaluation of recommender models

All work was implemented in Python using the SurPRISE package.

Dataset

We used the MovieLens-Ratings.csv dataset, which contains:

• userId: Unique user identifier
• movieId: Unique movie identifier
• rating: Explicit movie rating (0.5 to 5.0)
• timestamp: Seconds since Jan 1, 1970 (not used in the model)

The dataset includes over 27 million ratings and was pre-sorted by userId and movieId.

Contribution

My contribution includes data distribution by timestamp in Part 1, matrix factorization-based method and model comparison in Part 2.