Skip to content

Latest commit

 

History

History
58 lines (36 loc) · 2.5 KB

README.md

File metadata and controls

58 lines (36 loc) · 2.5 KB

Movie Recommender

Authors

Website link: Movie Recommender

Project Overview

This project involves creating a movie recommender website using Python and Streamlit. The recommender system is divided into two main systems:

  1. System 1 - Genre-based Recommender
  2. System 2 - Collaborative Filtering Recommender

System 1: Genre-based Recommender

The genre-based recommender system suggests movies based on the selected genre. The top movies in each genre are determined based on a weighted average rating.

Weighted Rating Formula

Example Image

Where:

  • mean_ratings is the mean of the ratings for the movie.
  • count is the number of reviews for this movie.
  • median is the median number of reviews for movies in this genre.

This formula helps in dealing with movies that receive only a few high-point reviews.

Implementation Overview

  1. Data Loading: Movies and ratings data are loaded from .dat files.
  2. Data Preprocessing: The data is merged and aggregated to compute mean ratings, counts, and median counts.
  3. Weighted Ratings Calculation: The weighted average ratings are computed for each movie.
  4. Genre Selection: Users can select a genre from a predefined list.
  5. Display Top Movies: The top 10 movies in the selected genre are displayed with their images and ratings.

System 2: Collaborative Filtering Recommender

The collaborative filtering system suggests movies based on user ratings. Users rate some movies, and recommendations are generated using Item-Based Collaborative Filtering (IBCF).

Implementation Overview

  1. Data Loading: Movies data is loaded.
  2. User Rating Collection: Users rate a set of displayed movies.
  3. Generate Recommendations: Using the user's ratings, the system generates a list of recommended movies.
  4. Display Recommendations: Recommended movies are displayed with their titles and genres.

Conclusion

The Movie Recommender Website provides two different recommendation systems:

  • Genre-based Recommender: Suggests movies based on selected genres using weighted average ratings.
  • Collaborative Filtering Recommender: Provides personalized movie recommendations based on user ratings.

Both systems leverage Streamlit for an interactive and user-friendly web interface, enabling users to easily discover new movies based on their preferences and ratings.