🎬 Movie Recommendation System – Content-Based & Collaborative Filtering

1. Problem Statement and Goal of Project

The goal of this project is to build a movie recommendation system using two core approaches:

1. Content-Based Filtering – Recommending movies similar to those a user already likes, based on movie attributes.
2. Collaborative Filtering – Recommending movies based on the preferences of users with similar tastes.

The project also explores the advantages and limitations of each method, with an intention to potentially combine them in a future hybrid recommender system.

---

2. Solution Approach

A. Content-Based Filtering

• Feature Extraction

  • Extract movie release years from titles.
  • Split the genres column into individual genres.
  • Apply One-Hot Encoding to create binary genre columns (0 = not in genre, 1 = in genre).

• User Profile Building

  • Introduce an example user with predefined ratings.
  • Identify genres of rated movies and compute weighted scores for each genre based on user ratings.

• Recommendation Generation

  • Compare all movies’ genre vectors against the user’s weighted genre profile.
  • Rank movies by similarity score and recommend the top results.

• Advantages & Limitations

  • Strengths: Personalization, reduced cold start for new users, transparency in recommendations.
  • Weaknesses: Over-specialization, limited diversity, dependency on feature quality.

---

B. Collaborative Filtering

• User-Based Collaborative Filtering

  • Find other users with similar rating patterns to the target user using historical ratings.
  • Recommend movies that similar users have rated highly and the target user hasn’t seen.

• Advantages & Limitations

  • Strengths: Can suggest unexpected items outside user’s known preferences, adaptable to changing tastes.
  • Weaknesses: Cold start problem, data sparsity, scalability issues, potential privacy concerns.

---

3. Technologies & Libraries

• Python
• pandas – Data manipulation and preprocessing
• NumPy – Numerical operations for profile and similarity calculations
• Matplotlib – Visualization of results (inline in Jupyter)

---

4. Description about Dataset

• movies.csv – Contains movieId, title, and genres for each movie.
• ratings.csv – Contains userId, movieId, rating, and timestamp.
• Timestamp is dropped for this implementation as it is not currently used in recommendation logic.

---

5. Installation & Execution Guide

1. Clone the repository:

   git clone <repo-url>
   cd <repo-folder>

2. Install dependencies:

   pip install pandas numpy matplotlib

3. Run the notebook:

   jupyter notebook "recommendation system project.ipynb"

---

6. Key Results / Performance

• Content-Based Filtering successfully recommends movies based on genre similarity to user preferences.
• Collaborative Filtering provides recommendations based on similar users’ ratings.
• The notebook contains step-by-step implementation and comparison of both methods.

---

7. Screenshots / Sample Outputs

Typical outputs in the notebook include:

• One-Hot Encoded genre matrix.
• Weighted genre profile for a sample user.
• Ranked list of recommended movies with highest similarity scores.
• Collaborative filtering recommendation lists.

---

8. Additional Learnings / Reflections

• Demonstrates a clear understanding of both content-based and collaborative filtering principles.
• Shows end-to-end preprocessing, feature engineering, and recommendation scoring without relying on pre-built recommender libraries.
• Highlights the trade-offs between the two methods and opens the path for future hybrid recommender systems.

---

👤 Author

Mehran Asgari 📧 [email protected] 🌐 https://github.com/imehranasgari

---

📄 License

This project is licensed under the Apache 2.0 License – see the LICENSE file for details.

---

💡 Some interactive outputs (e.g., plots, widgets) may not display correctly on GitHub. If so, please view this notebook via nbviewer.org for full rendering.

---