🎥 Netflix-Style Movie Recommender

A content-based movie recommender system built with Flask, combining machine learning and interactive visualizations.
Inspired by Netflix, this web app offers personalized recommendations, trending titles, and direct trailer links — all in a simple, clean interface.

---

✨ Features at a Glance

• 🔐 User Authentication: Sign up, log in, manage sessions securely.
• 🎯 Personalized Recommendations: Suggests similar movies using NLP-based content filtering.
• 📈 Trending Section: Highlights top-rated movies (rating > 9).
• 📊 Visualizations: Language distribution & genre-wise average rating charts.
• 🎬 Watch Trailers: One-click trailer links from YouTube.
• 🛒 Subscription Plans: Simulated subscription page with confirmation screen.

---

🧠 How It Works

This project uses a content-based filtering approach to recommend movies. Here’s an overview of the data flow and logic:

1️⃣ Data Collection & Cleaning

• Raw movie data is stored in dataset/movies_dataset.csv.

• clean_data.py:

  • Removes duplicates and incomplete records.
  • Standardizes text fields (lowercase, trims spaces).
  • Ensures all movies have valid ratings and image references.

• Clean data is saved to dataset/cleaned_movies.csv.

2️⃣ Feature Engineering & Model Training

• train.py:

  • Merges key metadata fields: tags, genre, actor, and language into a single text field.
  • Uses CountVectorizer (from scikit-learn) to create vector embeddings of each movie.
  • Computes cosine similarity between all movie vectors.

• Saves:

  • pkl/movies.pkl: DataFrame of cleaned movies with metadata.
  • pkl/similarity.pkl: Pre-computed similarity matrix.

3️⃣ Web Application (Flask)

• app.py:

  • Handles routing, user sessions, and rendering templates.
  • Supports login, signup, logout, recommendations, admin view, and data visualizations.

• Recommendations are served instantly using the pre-computed similarity matrix.

4️⃣ Database

• User accounts stored in SQLite (database/users.db).

• Created using netflixdb.py script with fields:

  • id (primary key)
  • email (unique)
  • password (plain text)

---

🏛 Project Structure

Netflix-App/
│
├── scripts/
│   ├── clean_data.py
│   ├── netflixdb.py
│   └── train.py
│
├── static/
│   ├── css/
│   │   ├── login.css
│   │   ├── signup.css
│   │   ├── style.css
│   │   ├── subscription.css
│   │   └── watch.css
│   │
│   └── img/
│
├── templates/
│   ├── admin.html
│   ├── home.html
│   ├── login.html
│   ├── payment_success.html
│   ├── signup.html
│   ├── subscription.html
│   ├── visualize.html
│   └── watch.html
│
├── dataset/
│   ├── movies_dataset.csv
│   ├── movie_links.csv
│   └── cleaned_movies.csv
│
├── pkl/
│   ├── movies.pkl
│   └── similarity.pkl
│
├── database/
│   └── users.db
│
└── app.py

---

⚙️ Main Modules & Scripts

Script	Purpose
clean_data.py	Cleans raw data, standardizes text, removes duplicates, handles missing values
train.py	Generates vectors & similarity matrix using CountVectorizer & Cosine Similarity
netflixdb.py	Creates SQLite database for user login
app.py	Flask app: handles routing, sessions, recommendations, visualizations

---

📊 Key Functionalities & Visualizations

• Visualizations (/visualize route)

  • Pie Chart : Shows distribution of movies by language
  • Bar Chart : Displays genres where the average rating exceeds 8
  • Dynamically generated using matplotlib and embedded directly in the web interface

• Watch trailers (/watch/<movie> route)

  • Opens the trailer/watch link using pre-collected YouTube URLs stored in movie_links.csv

• Subscription & Payment

  • /subscriptions : Displays sample subscription plans.
  • /payment_success : Confirms selected plan after form submission.

• Admin View (/admin)

  • Opens the login page. If logged in as admin, redirects to the homepage like a regular user.

---

🎬 Recommendation Logic

When a user selects a movie:

• The app retrieves its index from the movies DataFrame.
• Then extracts its similarity scores from the matrix.
• Sorts other movies by descending similarity score.
• Recommends the top 5 most similar movies.

Recommendations include:

• Title
• Genre
• Language
• Rating
• Poster image (via image_file column)

---

🧪 Technologies & Libraries Used

• Python & Flask: Web framework & backend
• SQLite: Lightweight relational database
• Pandas: Data cleaning & processing
• scikit-learn: NLP vectorization & similarity calculation
• Matplotlib: Data visualization
• HTML: Templates
• pickle: Used to save and load preprocessed data (movie metadata & similarity matrix)

---

🚀 How to Run Locally

• Navigate to your local Netflix-App folder, and open Command Prompt in that folder. Create a virtual environment :

  python -m venv venv  
  

• Activate the virtual environment :

  venv\Scripts\activate
  

• Install required dependencies : flask, pandas, scikit-learn, matplotlib.

  pip install flask pandas scikit-learn matplotlib
  

• Prepare the database :

  python scripts/netflixdb.py
  

  This creates users.db to store usernames and passwords.

• Clean the data :

  python scripts/clean_data.py
  

  This generates the cleaned_movies.csv dataset.

• Train the Recommendation Model :

  python scripts/train.py
  

  This generates movies.pkl & similarity.pkl.

• Run the app :

  python app.py
  

• Open your browser and go to :

  http://127.0.0.1:5000
  

---

📸 Preview Images (Screenshots of the App)

• Homepage :

• Recommendations Page :

• Visualizations Page :

• Sign-up Page :

• Sign-in Page :

• "Watch Now" Page :

• Subscriptions Page :

• "Payment Success" Page :

---

🚀 Possible Future Improvements

• Replace plain-text password storage with secure hashing

• Allow users to rate movies and track favorites or watchlists

• Add search auto-suggestions with fuzzy matching for better UX

• Integrate interactive visualizations using Plotly or Chart.js

• Enable basic subscription logic (e.g., restrict features based on plan)

• Improve UI with animations and responsive layouts

• Add movie filtering by genre, language, or rating

---

👤 Author

Built by Avik — For learning, experimenting, and exploring Machine Learning with real-world web apps.

---

📄 License

This is an open-source portfolio project. Feel free to use, modify, or extend it!

---