A Visual Analytics System for Exploring, Monitoring, and Forecasting Road Traffic Congestion

Area of Science:

• Computer Science
• Transportation Engineering
• Data Science

Background:

• Traffic congestion is a major urban challenge impacting efficiency and quality of life.
• Existing systems often lack integrated capabilities for exploration, surveillance, and accurate forecasting of congestion.

Purpose of the Study:

• To develop an interactive visual analytics system for exploring, surveilling, and forecasting traffic congestion.
• To identify causes and propagation directions of traffic congestion.
• To improve traffic management and congestion broadcasting.

Main Methods:

• Developed an interactive visual analytics system incorporating domain expert collaboration.
• Integrated a Long Short-Term Memory (LSTM) model for traffic congestion forecasting.
• Designed a weighting method to detect congestion causes and propagation directions.
• Utilized Volume-Speed Rivers (VSRivers) visualization for presenting traffic volume and speed data.

Main Results:

• The developed LSTM model demonstrated higher accuracy compared to other forecasting algorithms.
• The system effectively visualizes congestion causes, directions, and severity.
• Case studies and expert feedback confirmed the system's utility in traffic management and broadcasting.

Conclusions:

• The visual analytics system provides an effective tool for understanding and managing traffic congestion.
• The integration of LSTM forecasting and novel visualization enhances traffic data analysis.
• The system shows significant potential for real-world applications in urban traffic management.