Hybrid AI Movie Recommendation System for Pre-2015 Films

Mohammad

Mohammad Umar

Hybrid AI Movie Recommendation System

Award-winning ML project for Hollywood films released before 2015

πŸ† Project Highlights

πŸ₯‡ Best BSc Project (Grade A) β€” Selected among top 1% of cohort
🎯 2 Competition Finalist β€” Showcased at:
Project Expo Chandigarh University 2025 (Finalist, offered direct interview by one of the judges)
National Science Day 2025 (Finalist)
πŸŽ“ Built with: Python, Streamlit, Scikit-learn, Pandas, MovieLens 20M
πŸ•°οΈ Optimized for films released before 2015 (when user ratings are more consistent)

πŸ“– Overview

This hybrid movie recommender system blends both collaborative and content-based filtering to recommend relevant Hollywood films.
It was trained on over 27,000 movies and 20 million user ratings, using:
πŸ”— KNN-based collaborative filtering for taste matching
🧠 TF-IDF + Cosine similarity for genre-based similarity
βš–οΈ Hybrid weighting to balance both approaches
β€œParticularly excels with pre-2015 classic cinema, where user behavior patterns are strongest.”

🧠 System Architecture

Loading
graph LR
A[User Input] --> B(Title Matching)
B --> C{Hybrid Engine}
C --> D[KNN]
C --> E[Cosine Sim]
D --> F[Recommendations]
E --> F

πŸ“Š Model Performance

Metric Score RMSE 0.87 Coverage 94% (pre-2015) Diversity 0.68

πŸš€ How to Run Locally

git clone https://github.com/UmarTests/AI-Movie-Recommendation-System.git
cd AI-Movie-Recommendation-System
pip install -r requirements.txt
streamlit run Movei_rec_app.py

πŸ›  Key Features

🎯 Personalized suggestions using hybrid ML techniques
πŸ—ƒοΈ Trained on MovieLens 20M dataset
🧩 Robust against title misspellings
πŸ–₯️ Optional Streamlit UI for end-user interaction
πŸŽ₯ Focused on films released before 2015 for rating strength

πŸŽ“ Academic Recognition

πŸ“Œ Capstone Project for BSc in Computer Science
πŸ… Grade A and Best Project Award
πŸ† Finalist and awardee in multiple competitions
🧠 Demonstrated core ML techniques: matrix factorization, sparse vectors, cosine distance

✨ Why This Project Stands Out

+ Demonstrates mastery in applied ML and recommender systems  
+ Focused on pre-2015 cinema: real-world data quirks handled
+ End-to-end ready: From notebook β†’ deployable UI
+ Git versioned, modular Python architecture

πŸ“‚ File Structure

β”œβ”€β”€ Movie_Rec_proj.py           # Recommender engine logic  
β”œβ”€β”€ Movei_rec_app.py # Streamlit UI (optional)
β”œβ”€β”€ recommendation_system.py # Core similarity functions
β”œβ”€β”€ requirements.txt # All dependencies
β”œβ”€β”€ README.md # You're reading it

πŸ“« Let’s Connect

🌐 Portfolio : https://github.com/UmarTests

πŸ“„ License

MIT License – Free to use, fork, and learn from.
For detailed methodology and dataset processing, see the πŸ“„ Project Report (PDF)

🌐 INTRO LANDING PAGE (For GitHub Pages or index.html)

<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<title>Hybrid Movie Recommendation System</title>
<style>
body { font-family: sans-serif; background: #111; color: #eee; padding: 2rem; text-align: center; }
h1 { font-size: 2.5rem; }
p { font-size: 1.2rem; }
.gif { margin-top: 2rem; }
a.button {
display: inline-block; padding: 0.8rem 1.5rem; background: crimson;
color: white; text-decoration: none; border-radius: 8px; margin-top: 1rem;
}
</style>
</head>
<body>
<h1>🎬 Hybrid Movie Recommendation System</h1>
<p>An award-winning AI engine for recommending movies released before 2015.</p>
<p><strong>Built with:</strong> Python, Streamlit, scikit-learn, MovieLens</p>

<div class="gif">
<img src="https://media.giphy.com/media/3o7TKsrfldmm8cukWs/giphy.gif" width="480" />
</div>

<a href="https://github.com/UmarTests/AI-Movie-Recommendation-System" class="button">πŸ”— View on GitHub</a>
</body>
</html>
Like this project

Posted Aug 30, 2025

Developed a hybrid AI movie recommendation system for pre-2015 films.