This project is a Flask-based web application that recommends and displays films through various interfaces and filtering options. It interacts with a collection of data files downloaded from Kaggle, processes them, and then uses both content-based and collaborative filtering methods to generate film recommendations. The application also allows you to view random selections of films, filter by genre, and view top-rated lists.
- Features
- Front End / Back End
- Tech Stack
- Prerequisites
- Installation
- Project Structure
- App Screenshots
- Usage
- Notes
- Contributions
- Author
- Top-Rated Films: Displays a curated list of films based on weighted rating scores.
- Genre-Based Filtering: Lets you select a genre (such as action) and fetches films that belong to that category.
- Collaborative + Content-Based Filtering: Combines user rating data with film metadata to produce improved recommendations (via SVD from the surprise library).
- Random Selection: Fetches a random set of film posters with details like release date and production companies.
- Interactive Web Interface: Built with Flask, features dynamic routes and a JavaScript front end for toggling between poster images and film details.
- Uses HTML, CSS and Javascript to present film information and styling on the user-facing side.
- Contains clickable film posters that reveal more information upon flipping.
- JavaScript (script.js) manages DOM manipulation, toggling between poster images and additional film details, and controlling which buttons are active.
- Based on Python 3.12+ and the Flask framework, handling routes (/home, /rating, /genre, /similar) and query parameters.
- Employs pandas and scikit-learn for data wrangling and content-based filtering (cosine similarity).
- Uses surprise for training and loading the SVD model to incorporate collaborative filtering.
- Python 3.12+
- Flask – Web framework
- pandas - Data handling and analysis
- scikit-learn – Cosine similarity for content-based filtering
- surprise – Collaborative filtering (SVD model)
- requests – Checking poster validity
- HTML/CSS/JavaScript – For the web front end
- Python 3.12 or above installed on your machine.
- Kaggle API credentials if you wish to download the dataset automatically. Otherwise, place the “movies dataset” files in the correct location manually.
- Clone this Repository:
git clone /~https://github.com/hub-mm/movie_recommendation_system.git
- Navigate to project folder:
cd movie_recommendation - Install required Python packages:
pip install -r requirements.txt
- (Optional) Download Kaggle dataset:
- Ensure your Kaggle CLI is set up and your kaggle.json credentials are correctly configured.
- The script data_download.py will automatically fetch the dataset and place it in ./data/movies_dataset/7/ if not already present.
.
├── README.md
├── app
│ ├── static
│ │ ├── script.js
│ │ └── style.css
│ ├── templates
│ │ ├── genre.html
│ │ ├── index.html
│ │ ├── new_user.html
│ │ ├── rating.html
│ │ ├── sign_in.html
│ │ ├── similar.html
│ │ └── user_page.html
│ └── web_app.py
├── config.py
├── data
│ ├── movies_dataset
│ │ └── 7
│ │ ├── credits.csv
│ │ ├── keywords.csv
│ │ ├── links.csv
│ │ ├── links_small.csv
│ │ ├── movies_metadata.csv
│ │ ├── ratings.csv
│ │ └── ratings_small.csv
│ └── user_built
│ ├── cosine_sim.pkl
│ ├── count_vectorizer.pkl
│ ├── indices.pkl
│ ├── preprocessed_df.pkl
│ └── preprocessed_df_movies.pkl
├── main.py
├── model
│ └── model_filtering.pkl
├── requirements.txt
└── scripts
├── build_model
│ └── build_model_collab_filtering.py
├── data_management
│ ├── data_download.py
│ ├── data_merge_hybrid.py
│ ├── data_merge_similar.py
│ ├── data_preprocess_credits.py
│ ├── data_preprocess_keywords.py
│ ├── data_preprocess_links.py
│ ├── data_preprocess_movies_metadata.py
│ ├── data_preprocess_movies_metadata_posters.py
│ └── data_preprocess_ratings.py
├── script_app
│ ├── random_image.py
│ └── script_users.py
└── top_chart
├── top_chart_genre.py
├── top_chart_hybrid.py
├── top_chart_rating.py
└── top_chart_similar.py- app/web_app.py: Main Flask application file defining the routes.
- app/templates/*.html: HTML templates for each route (home, rating, genre, similar).
- app/static/: Contains JavaScript and CSS for the front end.
- scripts/data_management/: Scripts to download, preprocess, and merge datasets.
- scripts/top_chart/: Scripts generating top charts or recommendations based on various criteria.
- scripts/script_app/random_image.py: Provides random film posters.
- scripts/script_app/script_users.py: User class.
- model/: Stores the pickled collaborative filtering model.
Some film images arent available due to possible old link.
Home menu filters through images to try minimise redundant links.
- Run the Flask App:
python -m app.web_app
- Access the Web Interface:
- By default, navigate to: http://localhost:8000/home
- Use the buttons in the menu to switch pages:
- Home - random selection
- Genre - filter by genre
- Rating - top-rated films
- Similar - find similar films using hybrid filtering
- Login and Create User - log in to start session and have user specific info
- User Page - page with user specific favourite films
- Adjust Parameters:
- Use the query parameters in the URL or the interface’s text input to control the number of films displayed, the genre choice, or the film title for similarity.
- If the dataset or model is not found locally, the first run may attempt to download and preprocess everything, which could take a while.
- Some scripts (e.g., build_model_filtering.py) train or reload the collaborative filtering model. Ensure you have sufficient system resources (CPU/RAM).
Contributions, suggestions, and improvements are welcome. Feel free to open an issue or submit a pull request.
Created by hub-mm. For queries, feel free to reach out or open a GitHub issue. Enjoy discovering new films!