Farrelly, Gaba, & Kalala Mutombo - Hands-On Network Science | PyData Global 2023

Network science provides computationally feasible approaches for analyzing complex data relationships compared to traditional parametric models

Key network analysis tools include NetworkX and iGraph, with Python being recommended for beginners

Networks can effectively model various real-world scenarios:

• Social networks and relationships
• Disease spread and epidemics
• Stock market behavior and volatility
• Food pricing and supply chains
• Geographic relationships

Network analysis benefits include:

• Easy visualization of relationships
• Interpretable results
• Scalable computations
• Less parameter fitting required vs traditional models

The Shape of Data and Structure of Complex Network Theory are recommended starting resources for learning network science fundamentals

Network structures feature important elements:

• Hubs (densely connected regions)
• Bridges (connections between different groups)
• Weighted vs unweighted connections
• Thresholded relationships

Network metrics used for analysis include:

• Forman-Ricci curvature
• Betweenness centrality
• Page rank centrality
• Degree centrality

Time series analysis can be enhanced through network approaches:

• Using windowed correlation analysis
• Detecting change points
• Identifying market volatility periods

Network science has applications in epidemic control through:

• Modeling disease spread
• Identifying critical nodes for intervention
• Testing containment strategies

Future directions include expanding applications in spatio-temporal analysis and scaling solutions for larger datasets