Movie Recommendation System

This document describes a movie recommendation system designed to help users discover movies based on their preferences.

Link:

https://movie-recommend.azurewebsites.net

Technologies and Frameworks

• Flask
• Azure

Dataset

The system leverages data from the IMDb: https://www.imdb.com/ movie database through two phases:

Phase 1: GroupLens Data

• Data source: GroupLens: https://grouplens.org/datasets/
• Files used:
  • movies.csv (movieId, title, genre)
  • links.csv (movieId, imdbId, tmdbId)
  • ratings.csv (userId, movieId, rating)
• Preprocessing:
  • Removed unused files (genome-*.csv, tags.csv)
  • Filtered ratings.csv:
    • Removed users with less than 100 ratings.
    • Removed movies with less than 100 ratings.
    • Dropped userId column.
    • Calculated average rating per movie.

Phase 2: Cinemagoer Library

• Fetches additional movie attributes using imdbId:
  • plot
  • cast
  • crew
  • director
  • countries
  • languages
  • production companies

Pipeline

The system follows a sequential pipeline:

1. Data Ingestion:

• Downloads data from GitHub.
• Unzips and saves data to DataIngestionArtiacts folder.

2. Data Preprocessing:

• Cleans and prepares data.
• Generates two CSV files:
  • movieId, title, genres, imdbId (saved in DataPreprocessingArtiacts)
  • imdbId (used for further data collection)

3. Data Collection:

• Fetches additional attributes using Cinemagoer library and imdbId.
• Retrieves data from GitHub by default (faster).
• Option to fetch live data by setting COLLECTION_FLAG to True (slower).
• Saves data to DataCollectionArtiacts folder.

4. Data Transformation:

• Combines data from both phases.
• Transforms data for model development.
• Saves data to DataTransformationArtiacts folder.

5. Model Development:

• Calculates movie similarities using CountVectorizer and cosine similarity.
• Uses title and imdbId for recommendations.
• Saves model artifacts to ModelDevelopmentArtiacts folder (including a Data folder).

6. Website Development:

• Flask framework is used to create a user interface.
• Users can select a movie and receive recommendations.

Execution:

To execute this project and run the pipeline, install all the requirements and run the following command:

 	python main.py   

Execution of the pipeline can be seen in the logs folder that will create during the run. All the components are saved inside the artifacts folder. A Data folder is also created that uses in the web development part.

Demo of Project:

Make sure you have executed main.py file. For demo of the project can be done by running the following command:

	python app.py

Deployment:

This project is deployed on the Azure, and this is continuos deployment.

Link:

https://movie-recommend.azurewebsites.net/

Contributor:

• Ravi Kumar