Urban Street Network Analysis and Traffic Safety

Overview

This project conducts comprehensive street network analysis across multiple US cities using advanced network science techniques. At its core, the project implements Multiple Centrality Assessment (MCA) to analyze urban street patterns, then correlates these network properties with geospatial traffic accident data. By examining how network structure relates to traffic safety outcomes, this research provides insights into the relationship between urban design and public safety.

Features

• Multi-city street network extraction and processing using OSMnx
• Comprehensive network topology metrics calculation (redundancy index, gamma index, cyclomatic number)
• Multiple Centrality Assessment (MCA) analyzing three complementary measures:
  • Degree centrality (local connectivity at intersections)
  • Betweenness centrality (importance in facilitating movement across the network)
  • Closeness centrality (global accessibility and network efficiency)
• K-means clustering to identify distinct urban patterns based on centrality profiles
• Integration with geospatial US traffic accident data
• Spatial correlation analysis between network properties and accident distributions
• Advanced visualization techniques for network metrics and clusters

Advantages of Multiple Centrality Assessment (MCA)

MCA provides significant advantages over single-metric analyses:

• Comprehensive understanding: Captures different aspects of network functionality that single metrics miss
• Complementary perspectives: Combines local (degree), global (closeness), and intermediary (betweenness) measures
• Pattern identification: Reveals structural patterns that would be invisible to individual metrics
• Robust clustering: Enables more meaningful classification of urban areas
• Better correlation with real-world phenomena: Provides stronger explanatory power for spatial patterns of traffic accidents
• Decision support: Offers actionable insights for urban planning and traffic safety interventions

Requirements

• Python 3.x
• Required packages:
  • geopandas
  • osmnx
  • networkx
  • pandas
  • numpy
  • scikit-learn
  • matplotlib
  • seaborn
  • tqdm
  • pyproj

Installation

pip install osmnx geopandas networkx pandas numpy scikit-learn matplotlib seaborn tqdm pyproj

Usage

The project workflow follows these steps:

1. Network Extraction: Extract street networks for selected US cities
2. Network Metrics: Calculate basic topology statistics
3. Multiple Centrality Assessment: Compute and analyze degree, betweenness, and closeness centrality
4. Centrality Visualization: Visualize the distribution of each centrality measure
5. Cluster Analysis: Apply K-means clustering based on multiple centrality measures
6. Accident Data Integration: Map traffic accident data to the street network
7. Correlation Analysis: Study relationships between network properties and accident patterns
8. Results Visualization: Create visual representations of findings

Key Findings

The research reveals how different street network structures correlate with traffic safety:

• Certain centrality clusters show statistically significant relationships with accident rates
• Global network properties can help predict local safety patterns
• Inter-city comparisons reveal how different urban designs influence safety outcomes

Example Applications

• Identifying high-risk street configurations
• Prioritizing safety interventions based on network properties
• Informing urban planning decisions with safety considerations
• Comparative analysis of city designs and their safety implications

Future Work

• Temporal analysis of accident patterns in relation to network evolution
• Expansion to additional cities and international comparisons
• Integration with other urban data (demographics, land use, traffic volume)
• Development of predictive models for accident risk based on network properties
• Creation of network-based safety indices for urban areas

Data Sources

• Street network data: OpenStreetMap via OSMnx
• Accident data: US Accidents dataset (March 2023)

License

[Your chosen license]

Acknowledgments

• OpenStreetMap contributors
• OSMnx library by Geoff Boeing
• Contributors to the US Accidents dataset