🎬 Netflix Movie Recommendation System

A content-based movie recommendation system built using Python, Pandas, and scikit-learn, designed to suggest movies/TV shows based on similarity in metadata such as language, availability, release date, and viewing statistics.

The system supports retraining on new datasets and provides a Flask-based web interface for real-time recommendations.

---

📌 Features

• Content-Based Filtering: Recommendations are based on similarity in categorical and numerical attributes (e.g., genre, language, popularity).
• Data Cleaning Pipeline: Automatic handling of missing values, duplicate titles, and inconsistent formats.
• Custom Feature Engineering: Generates Content_ID for efficient mapping and similarity computation.
• Web Interface: Flask app for user-friendly recommendation queries.
• Retraining Support: Easily retrain on updated or custom datasets.
• Optimized for Deployment: Saves model, preprocessor, and metadata for quick loading.

---

⚙️ How It Works

1. Data Loading & Cleaning

   • Reads CSV dataset
   • Cleans numeric fields (Hours Viewed)
   • Drops duplicate and missing titles
   • Extracts Release_Year from Release Date
   • Assigns unique Content_ID to each item

2. Feature Engineering

   • Categorical Columns: One-hot encoded
   • Numerical Columns: Standard scaled
   • Saves metadata and preprocessor for future use

3. Recommendation

   • Computes cosine similarity between feature vectors
   • Returns top-N most similar items to a given title

---

🚀 Getting Started

1️⃣ Clone the Repository

git clone https://github.com/yourusername/Netflix_Movie_Recommendation_System.git
cd Netflix_Movie_Recommendation_System

2️⃣ Prepare the Dataset

Place your dataset (CSV) inside the data/ folder.
Required Columns:
Title Available Globally? Release Date Hours Viewed Language Indicator Content Type

3️⃣ Retrain the Model (Optional)

If you want to train on your dataset:

python train.py --input (datapath) --outdir ./models --neighbors 50(optional)

4️⃣ Run the Web App

python app.py

Browser View:

http://127.0.0.1:5000/

🎯 Usage Example

Example Query:

Input: wednesday

Output: Similar shows/movies based on metadata (language, year, popularity, etc.).

The recommendations are not random — they are based on metadata similarity, meaning the system suggests titles with similar language, release period, and audience engagement patterns.

🔮 Future Improvements

• Include genre-based similarity from NLP on movie descriptions

• Add collaborative filtering using user ratings

• Support multi-language search

• Deploy the app to Heroku / Render

👨‍💻 Author

Dipean Dasgupta
Computer Science Graduate | EdgeAI & ML Enthusiast