Spotify Feature Collaboration Network Analysis

This project constructs and analyzes a Spotify artist collaboration network, containing ~156,000 artists and 300,000+ connections. Using network science techniques, it investigates small-world properties, power-law degree distribution, and community structures.

• Examines the giant component and subgraphs based on genres & country-specific charts.
• Applies Louvain community detection to identify collaboration clusters.
• Investigates genre distributions and co-occurrences among chart-topping artists.

This project utilizes the Spotify Artist Feature Collaboration Network dataset from Kaggle: https://www.kaggle.com/datasets/jfreyberg/spotify-artist-feature-collaboration-network

---

Research paper publication efforts in progress.

Author: Raquel Ana M Bush

Project begun for course: MTH 565 - Network Analysis (under Professor Dana Fine, University of Massachusetts Dartmouth)

---

Running the Notebook

1. Open the Jupyter Notebook:

jupyter notebook spotify_feature_network.ipynb

(If Jupyter Notebook is not installed, run: pip install notebook)

2. Install dependencies:

pip install pandas
pip install networkx
pip install numpy
pip install matplotlib
pip install scipy
pip install scikit-learn
pip install community
pip install py4cytoscape

3. Modify the following lines in the notebook to point to your local dataset paths:

nodes_path = "your_nodes.csv_file_path"
edges_path = "your_edges.csv_file_path"

4. Run the cells sequentially to:

• Load the dataset.
• Construct and analyze the graph and subgraphs.
• Generate visualizations of communities, distributions, clustering, and diameter properties.

5. Interpret the results from the plots and network statistics. I've included guidance in markdown throughout.