A two-phase clustering approach for traffic accident black spots identification: integrated GIS-based processing and...

Area of Science:

• Traffic Safety Engineering
• Geographic Information Systems (GIS)
• Data Mining

Background:

• Effective identification of road traffic accident black spots is crucial for enhancing road safety.
• Existing methods often face challenges in accuracy and parameter subjectivity.

Purpose of the Study:

• To propose a novel model for accurate black spot identification by integrating GIS with hierarchical density-based spatial clustering.
• To objectively determine optimal clustering parameters using the density-based clustering validation index.

Main Methods:

• Developed a model combining GIS processing with hierarchical density-based spatial clustering of applications with noise (HDBSCAN).
• Employed the density-based clustering validation index for optimal parameter selection, reducing subjectivity.
• Validated the model using 3536 accident data from Hangzhou, China (August-October 2020).

Main Results:

• Identified 39 black spots, which accounted for 75% of all accidents within only 23.26% of the total road network length.
• The proposed model outperformed conventional HDBSCAN and K-means models in concentrating accidents per unit road length.
• On-site surveys confirmed the high recognition accuracy of the identified black spots.

Conclusions:

• The integrated GIS and HDBSCAN model offers a robust and objective approach to black spot identification.
• The findings demonstrate the model's effectiveness in pinpointing high-risk road segments for targeted safety interventions.
• The identified black spots warrant further investigation and potential improvements to mitigate traffic accidents.