Movie Recommender System Project in Python

The Movie Recommender System Project in Python (CineMatch) is a complete machine learning and web development package built for final year submissions. It analyses a catalogue of nearly 4,800 movies and recommends the closest matches to any title you select, using content-based filtering rather than simple popularity lists. The application combines a Flask backend, a custom single-page frontend, and a full content-based recommendation pipeline. Everything runs locally on your machine with no API keys and no internet dependency.

Key Features

• Five separate recommendation engines deliver matches by overall similarity, shared genres, same production studio, similar themes, and shared cast.
• Detailed movie information is available for every title, including overview, rating, runtime, budget, revenue, spoken languages, director, and top cast.
• A modern hand-built interface uses a dark gradient theme with live search autocomplete, animated stat counters, gradient poster cards, and a details modal.
• The entire system works offline, with no TMDB API calls, no external image hosts, and no CDN dependencies during a demo.
• Smart caching stores the computed similarity matrices as pickle files, so the app loads almost instantly after the first run.
• A clean JSON API separates the Flask backend from the frontend, exposing tidy endpoints for movie lists, recommendations, and details.
• Content-based filtering is implemented with CountVectorizer and cosine similarity, both of which are strong viva discussion topics.
• Gradient placeholder cards are generated on the fly for movie posters, keeping the project fully self-contained.
• The complete package includes source code, a project report, a synopsis, and a PPT presentation for submission.

Technology Stack

System Requirements

• Python 3.8 or above installed on the system.
• pip for installing the packages listed in requirements.txt.
• Flask for running the local web server.
• pandas, scikit-learn, and NLTK for data processing and similarity calculation.
• Any modern browser such as Chrome, Edge, or Firefox.
• Around 500 MB of free disk space for the dataset and cached similarity matrices.

How It Works

The two TMDB CSV files are merged and cleaned, pulling together overview, genres, keywords, cast, crew, and studio details for each movie. Each film is then converted into a bag of words, with stemming applied and stop words removed, and the content is grouped into five separate feature sets. CountVectorizer turns these tags into numeric vectors, and cosine similarity scores every movie against every other movie in the catalogue. When you pick a title, the results are sorted by similarity score and the top unique matches from all five engines are displayed instantly on screen.

Perfect For

This project suits BCA, MCA, B.Tech CS/IT, M.Tech, and MBA students who need a final year project that covers both machine learning and full-stack web development. It is well suited for viva presentations because vectorization, cosine similarity, and content-based filtering are common examiner questions that you can answer with a live demo. It also works as a strong portfolio and interview project, since recommender systems power platforms like Netflix, Amazon, and Spotify.

Future Enhancements

The project can be extended with real movie posters from a bundled image pack, collaborative filtering that blends user ratings with content matches, user accounts with watchlists and a personalised feed, filters for release year, language and minimum rating, cloud deployment for online access, a feedback loop where users rate recommendations, and an expanded dataset with newer titles.

Watch Project Demo

Watch the full project walkthrough and setup tutorial on our YouTube channel.

▶ Watch on YouTube – Decode It

More Details & Full Documentation

For complete project details, source code, and setup guide, visit the official project page.

🔗 View Full Project Details – UpdateGadh