🎬 TMDb Movie Dataset Analysis & Visualization

This project was completed as part of the STAT288: Data Visualization course at the University of Bahrain.
It explores patterns in movie preferences, ratings, and genre characteristics using comprehensive visualization techniques.

👥 Team

• Ebrahim Juma Shakak Alsawan
• Sadeq Jaafar Ali Deyab
• Salman Wael Salman Ahmed

🎯 Objective

To detect patterns related to people's movie preferences, revenue trends, and how various factors (genres, budget, production variables) affect movie outcomes through effective data visualization.

📊 Dataset Overview

Source: TMDb 5000 Movie Dataset from Kaggle
Size: 4,803 movies with comprehensive metadata including:

• Movie details (budget, revenue, runtime, ratings)
• Cast and crew information (106,257 cast entries, 129,581 crew entries)
• Genre associations (12,160 genre mappings)
• Production details (countries, companies)

🔍 Key Findings

Genre Quality Patterns

• Best performing genres: War (45%), History (42%), Western (40%)
• Worst performing genre: Horror (8% good movies)
• Definition: "Good movies" = average rating ≥ 7.0 with ≥30 votes

The Niche Genre Theory

Discovered a logarithmic relationship between movie count and quality:

Good Movies % = 71.93 - 7.84 × ln(Movie Count)

• Niche genres (fewer movies) → Higher quality ratings
• Popular genres (more movies) → Lower average ratings

Notable Exceptions

• Drama: Outperforms expectations despite high movie count
• Horror: Underperforms significantly compared to moderate movie count

🛠 Tools & Techniques Used

• Python: Data cleaning, statistical analysis, correlation matrices
• Tableau: Interactive scatter plots and relationship visualizations
• Pandas: Data manipulation and cross-tabulation
• Matplotlib/Seaborn: Heat maps and bar charts

📈 Visualizations Created

1. Horizontal Bar Chart: Percentage of good movies by genre
2. Correlation Heat Map: Genre associations and co-occurrence patterns
3. Scatter Plot: Movie count vs. quality percentage with logarithmic trendline

🔗 Genre Correlation Insights

• Good genres tend to correlate positively with other good genres (History ↔ War)
• Bad genres cluster together (Horror ↔ Thriller ↔ Action)
• Drama acts as a "bridge genre" - correlates with quality genres while maintaining high volume

📁 Project Structure

├── movies.csv                    # Main movie dataset (4,803 movies)
├── movie_genre.csv              # Genre associations (12,160 mappings)
├── movie_cast.csv               # Cast information (106,257 entries)
├── movie_crew.csv               # Crew details (129,581 entries)
├── movie_country.csv            # Production countries (6,436 entries)
├── movie_company.csv            # Production companies (13,677 entries)
├── data_cleaning.py             # Data preprocessing and transformation
├── python_visualizations.py     # Statistical analysis and plotting
└── report.pdf                   # Complete analysis report

🎬 Storytelling Narrative

"Why Does Movie Quality Vary So Dramatically Between Genres?"

Our analysis reveals that genre quality differences stem from audience targeting strategies rather than inherent genre characteristics. Niche genres succeed through focused storytelling for specific audiences, while mass-market genres sacrifice quality for broader appeal.

---