🎬 CineMatch — Hybrid Movie Recommender System

A hybrid recommendation engine combining Content-Based Filtering and Collaborative Filtering techniques to deliver personalized movie suggestions using the IMDB Movies Dataset.

---

📌 Table of Contents

• Overview
• Dataset
• Project Structure
• Methodology
• EDA Highlights
• Recommendation Engine
• Tech Stack
• Getting Started
• Results

---

🧠 Overview

CineMatch is a hybrid movie recommender system built on the IMDB Movies Dataset. It combines:

• Content-Based Filtering — recommends movies similar to a given title based on genres, directors, and cast using TF-IDF vectorization and cosine similarity.
• User-Based Collaborative Filtering — leverages user rating patterns to surface movies liked by similar users.

The hybrid approach addresses the limitations of each method individually — content-only systems are too narrow, while collaborative systems suffer from cold-start problems.

---

📂 Dataset

Property	Detail
Source	Kaggle — IMDB Movie Ratings Dataset
File	movie_data.csv

Features

Feature	Description
director_name	Director of the movie
genres	Pipe-separated genre tags (e.g. Action|Drama)
actor_1_name	Primary actor
actor_2_name	Secondary actor
actor_3_name	Tertiary actor
movie_title	Official title
imdb_score	Rating on a 1–10 scale
num_voted_users	Number of IMDB user votes
title_year	Year of release
duration	Runtime in minutes
language	Primary language
country	Country of production

---

📁 Project Structure

├── Hybrid-Recommender-System.ipynb   # Main notebook
├── movie_data.csv                    # Dataset
└── README.md

---

🔬 Methodology

1. Data Cleaning & Preprocessing

• Separated numeric and categorical features
• Imputed missing numeric values with median
• Imputed missing categorical values with "Unknown"
• Stripped invisible characters and whitespace from movie titles
• Validated data ranges and categorical consistency

2. Exploratory Data Analysis

• Distribution of IMDB scores
• Average score and duration trends over time
• Genre popularity and average ratings by genre
• Most active actors by movie count
• Top 10 most-voted movies

3. Feature Engineering

Combined text-based features into a single features column:

imdb['features'] = (
    imdb['genres'] + " " +
    imdb['director_name'] + " " +
    imdb['actor_1_name'] + " " +
    imdb['actor_2_name'] + " " +
    imdb['actor_3_name']
)

4. Content-Based Filtering

• Applied TF-IDF Vectorization (max_features=3000, bigrams, min_df=2)
• Computed Cosine Similarity matrix across all movies
• Retrieved top-N most similar movies for any given title

---

📊 EDA Highlights

• 🎭 Drama is the most common genre, followed by Comedy and Thriller
• ⭐ IMDB scores are right-skewed — most movies cluster between 6.0 – 7.5
• 🎬 Robert De Niro and Nicolas Cage lead in movie appearances
• 📈 Average movie ratings slightly declined post-2000
• 🔗 Engagement metrics (votes, reviews) are strongly correlated with each other but only moderately with IMDB score — popularity ≠ quality

---

🎯 Recommendation Engine

get_recommendations("Avatar")

Output:

Recommended movies similar to Avatar:
1. Aliens
2. Titanic
3. Guardians of the Galaxy
4. Star Trek Into Darkness
...

The function:

1. Looks up the movie index
2. Retrieves its cosine similarity scores against all other movies
3. Returns the top 10 most similar titles (excluding itself)

---

🛠 Tech Stack

Library	Purpose
pandas	Data loading, cleaning, transformation
scikit-learn	TF-IDF vectorization, cosine similarity
matplotlib	Plotting
seaborn	Statistical visualizations

---

---

✅ Results

Method	Strength
Content-Based	Strong at finding stylistically similar films (same director, cast, genre)
Collaborative	Captures user taste patterns beyond surface metadata
Hybrid	Balances both — broader coverage, better cold-start handling

---

📬 Contact

Have suggestions or want to collaborate? Feel free to open an issue or reach out!

---

Built with ❤️ using Python & IMDB data