Vehicle Motion and Conflict Data for Freeway Diverging Area Crash Prediction

Twelve interchange diverging areas from two multilane freeways in China were analyzed using image recognition to extract 48 vehicle motion parameters and surrogate safety measures. The dataset, created by Feng Tang and last updated in April 2026, innovatively applies Extended Time-to-Collision to label longitudinal and lateral conflicts at 30-second intervals. It is a 9.5 KB Excel file released under a CC-BY-4.0 license.

Use Cases

• Training machine learning models for crash risk prediction based on vehicle motion parameters and surrogate safety measures.
• Analyzing spatiotemporal patterns of lateral versus longitudinal conflicts in freeway diverging areas.
• Identifying key risk factors for dangerous driving warning systems using SHAP interpretability analysis.
• Benchmarking predictive performance of models like Random Forest, Neural Network, SVM, and XGBoost on traffic conflict data.

Strengths

• Includes 48 distinct vehicle motion parameters and surrogate safety measures extracted via image recognition.
• Applies the validated Extended Time-to-Collision metric for labeling two-dimensional lateral conflicts.
• Data is structured at a refined 30-second interval for temporal analysis.

Limitations

• Row count is unknown, which may limit suitability assessment for large-scale modeling.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset is very small at 9.5 KB, indicating limited scope or a sample.

Provenance

Source

figshare

Collection Method

Data extracted via image recognition technology from video of 12 interchange diverging areas.

Time Range

Time intervals are at 30-second resolution; specific collection dates are not provided.

Freshness

Last updated 2026-04-29 17:54:56; freshness should be verified.

Geography

Two multilane freeways in China.