Trajectory data based freeway high-risk events prediction and its influencing factors analyses

Area of Science:

• Traffic Safety
• Transportation Engineering
• Data Science

Background:

• Road crashes cause significant global losses.
• Existing traffic data lacks micro-level detail for crash analysis.
• High-risk events are crucial indicators for proactive safety measures.

Purpose of the Study:

• Analyze relationships between traffic operations and crash risk.
• Utilize microscopic trajectory data for high-risk event identification.
• Develop accurate predictive models for traffic safety.

Main Methods:

• Used the HighD Dataset for German highway traffic.
• Identified high-risk events using Modified Time to Collision (MTTC < 2s).
• Developed and compared logistic regression models (standard, RELR, RPLR).

Main Results:

• The Random-Parameter Logistic Regression (RPLR) model showed superior fitness and prediction accuracy.
• Disturbed longitudinal and lateral traffic flow positively impacts high-risk events.
• Close following distances significantly contribute to high-risk event occurrence.
• RPLR models achieved 97% prediction accuracy for events 2 seconds in advance.

Conclusions:

• Microscopic traffic data and advanced models like RPLR enhance crash risk prediction.
• Understanding traffic dynamics is vital for implementing effective safety countermeasures.
• The findings support proactive traffic safety improvements and future applications.