Movie Recommendation System

A content-based movie recommendation system using TF-IDF Vectorization and Cosine Similarity to suggest similar movies based on user input.

---

Features

Content-Based Filtering – Recommends movies based on similar genres, keywords, and cast.
TF-IDF Vectorization – Converts movie descriptions into numerical feature vectors.
Cosine Similarity – Computes similarity scores between movies.
User Input Matching – Uses fuzzy matching to handle spelling errors in movie searches.

---

Installation

Clone the Repository

git clone https://github.com/yourusername/movie-recommendation.git
cd movie-recommendation

Create a Virtual Environment (Optional but Recommended)

python -m venv venv
source venv/bin/activate  # For macOS/Linux
venv\Scripts\activate  # For Windows

Install Dependencies

pip install -r requirements.txt

---

Dataset

• The dataset is stored in movies.csv.
• Ensure that the movies.csv file is placed in the data/ folder.
• If the dataset is missing, download it from Kaggle or use a sample dataset.

---

Usage

Run the script

python recommendation.py

Example Input

Enter your favorite movie: Inception

Example Output

Movies suggested for you:
1. Interstellar
2. The Prestige
3. Memento
...

---

Evaluation Metrics

After generating recommendations, the following metrics are computed:

• Precision@k
• Recall@k
• F1 Score@k
• Mean Reciprocal Rank (MRR)

The metrics are calculated by comparing the recommended movies against a set of relevant movies or a ground truth for a group of users.

---

Future Improvements

🔹 Deploy as a Flask/Streamlit Web App
🔹 Add Collaborative Filtering for hybrid recommendations
🔹 Improve search accuracy using Word Embeddings (BERT, Word2Vec)
🔹 Implement GPU Acceleration for large-scale similarity computations

---

License

This project is licensed under the MIT License - feel free to use, modify, and share!

---

Contact

Your Name: Sudiksha Rajavaram