Watch Market GNN Code Repository

This repository contains the code implementation for creating and analyzing a Graph Neural Network (GNN) based dataset of the luxury watch market. The dataset itself is hosted on Hugging Face.

Repository Structure

/code

Contains the core implementation of the Graph Neural Network:

• watch_gnn.py: Primary GNN implementation including:

  • Data preprocessing
  • Feature engineering
  • Network construction
  • Embedding generation
  • Graph structure creation

  This is an overview of what functionality has been in the watch_gnn.py :

  

/visualization

Contains analysis and visualization scripts:

• watch_analysis.py: Implements various visualizations including:
  • UMAP embeddings
  • t-SNE analysis
  • PCA visualization
  • Force-directed graph
  • Starburst visualization
  • Brand distribution analysis
  • Correlation studies

/images

Contains visualization outputs:

1. Brand distribution treemap
2. Feature correlation matrix
3. UMAP visualization
4. t-SNE analysis
5. PCA visualization
6. Force-directed graph
7. Starburst graph
8. Network architecture diagram

Root Files

• requirements.txt: Lists all Python dependencies
• Watches.csv: Original dataset file
• .gitignore: Specifies files and directories to be ignored by Git

Setup and Installation

1. Clone the repository:

git clone https://github.com/calicartels/watch-market-gnn-code.git
cd watch-market-gnn-code

2. Create and activate a virtual environment:

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

3. Install dependencies:

pip install -r requirements.txt

Usage:

1. For GNN model implementation:

python code/watch_gnn.py

2. For visualization and analysis:

python visualization/watch_analysis.py

Dataset Access:

The complete dataset is available on Hugging Face: TMVishnu/watch-market-gnn

License

This project is licensed under the Apache License 2.0

Note:

This complete code took me close to 6 hours to run without a GPU on a Macbook Air M3 with 16GB RAM. Wait for the complete code to finish running.