README_movie_recommender.md

Movie Recommendation System Using Content-Based Filtering

This repository hosts a content-based movie recommendation system built with Python. It uses metadata from a movie dataset to suggest similar movies based on features like genres, cast, crew, and keywords.

---

Model Details

• Model Type: Content-Based Recommendation System
• Technique Used: Cosine Similarity
• Libraries: Pandas, Scikit-learn, Numpy
• Dataset: TMDB 5000 Movie Dataset (or similar metadata-rich dataset)
• Task: Movie Recommendation based on content similarity

---

Usage

Installation

pip install pandas scikit-learn numpy

Running the Model

import pandas as pd
from sklearn.feature_extraction.text import CountVectorizer
from sklearn.metrics.pairwise import cosine_similarity

# Load dataset
movies = pd.read_csv('movies.csv')

# Combine relevant features into a single string
movies['combined_features'] = movies['genres'] + ' ' + movies['keywords'] + ' ' + movies['cast'] + ' ' + movies['crew']

# Vectorize features
vectorizer = CountVectorizer()
feature_vectors = vectorizer.fit_transform(movies['combined_features'])

# Compute similarity matrix
similarity = cosine_similarity(feature_vectors)

# Define recommendation function
def recommend(movie_name):
    movie_index = movies[movies['title'] == movie_name].index[0]
    distances = similarity[movie_index]
    movie_list = sorted(list(enumerate(distances)), reverse=True, key=lambda x: x[1])[1:6]
    for i in movie_list:
        print(movies.iloc[i[0]].title)
        
# Example usage
recommend("Inception")

---

Performance Metrics

This is a heuristic model and doesn't have standard ML performance metrics like accuracy or F1. Evaluation is subjective based on relevance and user satisfaction.

---

Dataset Details

The dataset includes the following fields:

• Title
• Genres
• Keywords
• Cast
• Crew

Preprocessing includes:

• Removing nulls and duplicates
• Parsing nested JSON fields into readable text
• Combining features for vectorization

---

Repository Structure

.
├── movies.csv                  # Dataset file
├── recommendation_system.ipynb  # Main notebook
├── README.md                   # Documentation file

---

Limitations

• Not personalized; recommendations are the same for all users.
• Doesn't account for user ratings or feedback.
• Limited by the richness and correctness of metadata in the dataset.

---

Contributing

Suggestions and improvements are welcome! Feel free to open issues or pull requests to help improve this project.

---