Collaborative Filtering for Movie Recommendations

This repository contains a Jupyter Notebook that implements a movie recommender system using collaborative filtering.

Technologies Used

• TensorFlow
• Keras
• Python

Dataset

The dataset used is a subset of the ml-small dataset from DataLens and consists of:

• movies.csv: Contains movie information (ID, title, genres).
• ratings.csv: Contains user ratings for movies, including timestamps.

Installation

To run this notebook, you need Python 3.x and the following dependencies:

pip install pandas numpy matplotlib tensorflow

Steps Involved

Data Preprocessing

• Load movie and rating data.
• Merge datasets and clean unnecessary columns.
• Include movie ratings of a new user.
• Normalize and structure the dataset for learning.

Data Visualization

• Histogram of rating count per user.
• Bar chart of rating distribution.

Learning algorithm

• Implementation of cost function with regularization.
• Learn model using a custom training loop with TensorFlow and Keras.

Movie Rating Predictions

• Generate movie predictions for the new user.
• Recommend top-rated movies based on predicted ratings.