🎬 FILMFLIX – Movie Recommendation System

A Machine Learning-powered Movie Recommendation System built using Content-Based Filtering,Count Vectorization, and Cosine Similarity to generate personalized movie recommendations.

The application features a Netflix-inspired user interface developed with Streamlit and provides movie recommendations along with posters and trailer previews.

📌 Project Overview

Recommendation systems power modern platforms such as Netflix, Amazon, Spotify, and YouTube by helping users discover relevant content.

This project implements a Content-Based Recommendation Engine that analyzes movie metadata and identifies movies with similar characteristics. When a user selects a movie, the system recommends the most relevant alternatives based on textual similarity.

The project demonstrates core Machine Learning concepts including:

• Data Preprocessing
• Feature Engineering
• Natural Language Processing (NLP)
• Vectorization Techniques
• Similarity Measurement
• Recommendation Systems
• Interactive ML Application Development

🚀 Features

✅ Recommend Top 5 Similar Movies

✅ Interactive Netflix-Style User Interface

✅ Movie Poster Display

✅ Trailer Integration

✅ Fast Similarity-Based Recommendations

✅ Offline Recommendation Generation

✅ Streamlit-Powered Web Application

🧠 Machine Learning Concepts Used

Content-Based Filtering

The recommendation engine suggests movies based on similarity between movie attributes rather than user ratings.

Feature Engineering

Important movie information including:

• Genres
• Keywords
• Cast
• Crew

is combined into a unified textual representation called tags.

Count Vectorization

Movie metadata is converted into numerical feature vectors using Count Vectorizer, which represents each movie based on the occurrence of important words and features. These vectors serve as the foundation for calculating content similarity and generating recommendations.

Cosine Similarity

Cosine Similarity is used to measure how closely related two movies are based on their vector representations.

Movies with higher similarity scores are recommended to the user.

⚙️ System Workflow

1. Load movie dataset.
2. Perform preprocessing and cleaning.
3. Combine relevant metadata into tags.
4. Convert text into numerical vectors.
5. Compute similarity matrix.
6. Select user movie input.
7. Retrieve top similar movies.
8. Display recommendations, posters, and trailers.

🛠️ Tech Stack

Programming Language

• Python

Libraries & Frameworks

• Pandas
• NumPy
• Scikit-learn
• Streamlit
• Pickle

Machine Learning Techniques

• Content-Based Filtering
• TF-IDF Vectorization
• Cosine Similarity

📂 Project Structure

FILMFLIX/
│
├── movie.py                 # Main Streamlit Application
├── model.py                 # Recommendation Model Creation
├── movies.pkl               # Processed Movie Dataset
├── similarity.pkl           # Similarity Matrix (Generated Locally)
├── posters/                 # Movie Poster Assets
├── screenshots/             # Application Screenshots
├── requirements.txt         # Project Dependencies
└── README.md

📊 Dataset

The project uses movie metadata containing:

• Movie Titles
• Genres
• Keywords
• Cast Information
• Crew Information

The recommendation engine utilizes approximately 5,000 movie records for generating recommendations.

▶️ Running the Project Locally

Clone Repository

git clone https://github.com/sreevyarao/P_1_RecommenderSystem.git

cd P_1_RecommenderSystem

Install Dependencies

pip install -r requirements.txt

Generate Similarity Matrix

python model.py

Launch Application

streamlit run movie.py

⚠️ Note

The file similarity.pkl is not included due to GitHub size limits.

- Run model.py to generate it locally.

📸 Application Screenshots

Home Screen

Displays movie selection interface and recommendation controls.

screenshots/Intro Screen.png

Recommendation Screen

Generates personalized movie recommendations.

screenshots/main page(movie selection & recommendations).png

Movie Details Screen

Displays posters and trailer previews.

screenshots/final page(poster+trailers).png

🎥 Project Demonstration

Demo Video: https://youtu.be/Fp1ctRTAKZQ

📈 Key Learning Outcomes

Through this project, I gained hands-on experience with:

• Recommendation System Design
• Machine Learning Pipelines
• Feature Engineering
• Data Preprocessing
• Vector Space Models
• Similarity Metrics
• Streamlit Application Development
• Model Integration into Real Applications

🌟 Future Scope

Hybrid Recommendation System

Upgrade the current recommendation engine into a Hybrid Recommendation System by combining:

Content-Based Filtering Collaborative Filtering

This would improve recommendation quality by incorporating both movie characteristics and user behavior patterns.

Personalized User Profiles

Implement user authentication and recommendation history tracking to provide customized recommendations.

TMDB API Integration

Fetch:

• Dynamic Posters
• Ratings
• Reviews
• Trending Movies
• Additional Metadata

in real time.

Regional Language Support

Expand recommendations to:

• Telugu Movies
• Hindi Movies
• Tamil Movies
• Other Regional Cinema

Recommendation Evaluation

Add recommendation quality metrics such as:

• Precision@K
• Recall@K
• NDCG

for systematic model evaluation.

Cloud Deployment

Deploy the application using Streamlit Cloud and integrate scalable backend services.

👨‍💻 Author

Sarode Sreevya Rao

B.Tech – Computer Science & Engineering (AI & Machine Learning)

GitHub: https://github.com/sreevyarao

LinkedIn: https://linkedin.com/in/s-sreevya-rao-79b8b9323

Portfolio: https://sreevyarao.github.io/Portfolio-website

⭐ If you found this project interesting, consider giving it a star.