Game Genre Network Analysis

Network analysis of Steam video game genres over time, exploring how game genres co-occur and evolve across different eras of gaming.

Overview

This project analyzes the Steam video game catalog to understand:

• Genre co-occurrence patterns: Which game genres tend to appear together
• Temporal evolution: How genre relationships have changed from 1995 to 2015+
• Network metrics: Centrality, clustering, and community structure of genre networks
• Predictive modeling: Regression analysis to understand factors influencing game success

The analysis combines Python-based data processing with Gephi for advanced network visualization.

Data Sources

The analysis uses Steam video game data containing:

• Game metadata (name, release date, developer, publisher)
• Genre classifications
• User ratings and playtime statistics
• Platform and pricing information

Primary datasets:

• steam.csv - Main Steam games dataset (~27,000 games)
• genres.csv - Expanded genre information (~76,000 records)

Requirements

Software

• Python 3.9+
• Gephi 0.10+ - Download from gephi.org
• Jupyter Notebook or JupyterLab

Python Dependencies

pip install -r requirements.txt

Key libraries:

• pandas, numpy - Data manipulation
• networkx - Network construction and analysis
• matplotlib, seaborn - Visualization
• scikit-learn - Machine learning/regression

Installation

1. Clone the repository:

   git clone https://github.com/yourusername/Game-Genre-Network-Analysis.git
   cd Game-Genre-Network-Analysis

2. Create a virtual environment (recommended):

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

   pip install -r requirements.txt

4. Install Gephi from gephi.org for network visualization.

Project Structure

Game-Genre-Network-Analysis/
├── README.md
├── LICENSE
├── requirements.txt
├── .gitignore
│
├── data/
│   ├── raw/                    # Original datasets
│   │   ├── steam.csv
│   │   └── genres.csv
│   └── processed/              # Cleaned/transformed data
│       └── genres-fixed.csv
│
├── notebooks/
│   ├── 01_data_exploration.ipynb      # Data cleaning and EDA
│   ├── 02_network_construction.ipynb  # Network building and analysis
│   └── 03_regression_analysis.ipynb   # Predictive modeling
│
├── networks/
│   ├── genre_networks/         # Genre-to-genre networks
│   │   ├── genres.gexf
│   │   ├── genres-fixed.gexf
│   │   └── genre_attributes.gexf
│   ├── bipartite/              # Game-genre bipartite networks
│   │   ├── steam_bipartite.gexf
│   │   ├── steam_bipartite-fixed.gexf
│   │   └── filtered_bipartite.gexf
│   └── temporal/               # Time-sliced networks
│       ├── 95_steam.gexf       # 1995
│       ├── 00_steam.gexf       # 2000
│       ├── 05_steam.gexf       # 2005
│       ├── 10_steam.gexf       # 2010
│       ├── 15_steam.gexf       # 2015
│       └── 95_05_steam.gexf    # 1995-2005 combined
│
├── gephi/
│   └── projects/               # Gephi project files
│       ├── Main.gephi          # Primary genre network analysis
│       ├── bipartite.gephi     # Game-genre bipartite analysis
│       ├── GenreEvolution.gephi    # Temporal analysis
│       ├── SteamEvolution.gephi    # Steam catalog evolution
│       ├── GenreAttributes.gephi   # Genre with node attributes
│       └── try.gephi           # Experimental workspace
│
└── outputs/
    ├── figures/                # Generated visualizations
    ├── visuals/                # Publication-ready exports
    ├── gephi_metrics/          # Gephi-computed metrics
    └── clustering_coeff/       # Clustering analysis reports

Workflow

Step 1: Data Exploration and Preprocessing (Python)

Run notebooks/01_data_exploration.ipynb:

• Load and clean the Steam games dataset
• Parse genre information
• Handle missing values and data quality issues
• Compute basic statistics

Step 2: Network Construction (Python)

Run notebooks/02_network_construction.ipynb:

• Build genre co-occurrence networks
• Create bipartite game-genre networks
• Generate time-sliced networks for temporal analysis
• Export networks as GEXF files for Gephi

Step 3: Regression Analysis (Python)

Run notebooks/03_regression_analysis.ipynb:

• Feature engineering from game metadata
• Train regression models (Ridge, Random Forest)
• Evaluate model performance
• Analyze feature importance

Step 4: Network Visualization (Gephi)

1. Open Gephi and import GEXF files from networks/
2. Apply layouts:
   • ForceAtlas2 for genre networks
   • Fruchterman-Reingold for bipartite networks
3. Compute network statistics:
   • Degree centrality
   • Betweenness centrality
   • Clustering coefficient
   • Modularity (community detection)
4. Style nodes by centrality/community
5. Export visualizations to outputs/visuals/

Gephi Projects

Project	Description
Main.gephi	Primary genre co-occurrence network with community detection
bipartite.gephi	Game-genre bipartite network showing connections
GenreEvolution.gephi	Temporal analysis of genre relationships
SteamEvolution.gephi	Evolution of the Steam catalog over time
GenreAttributes.gephi	Genre network with additional node attributes

Key Findings

• Action and Indie genres are the most central in the network
• Strong clustering between related genres (e.g., RPG-Adventure, FPS-Action)
• Genre diversity has increased significantly from 2005 onwards
• Community detection reveals distinct genre clusters

Network Metrics

The analysis computes:

• Degree Centrality: Most connected genres
• Betweenness Centrality: Genres that bridge different clusters
• Closeness Centrality: How quickly a genre can "reach" others
• Clustering Coefficient: Local clustering patterns
• Modularity: Community structure

Reproducing the Analysis

1. Ensure all data files are in data/raw/
2. Run notebooks in order: 01_ → 02_ → 03_
3. Import generated GEXF files into Gephi
4. Load Gephi project files from gephi/projects/ to see pre-configured analyses

License

MIT License - see LICENSE for details.

Author

Subash Pandey

Acknowledgments

• Steam data sourced from public APIs and datasets
• Network analysis powered by NetworkX and Gephi