A Two-Stage Sequential Framework for Traffic Accident Post-Impact Prediction Utilizing Real-Time Traffic, Weather, and Accident Data
Bibliographic record
Abstract
Detecting road accident impacts as promptly as possible is essential for intelligent traffic management systems. This paper presents a sequential two-stage framework for predicting the most congested traffic level that appears after an accident and the recovery time required for returning to the level of service that existed at the accident report time. As fewer accident characteristics are available at the report time, stage one models rely on real-time traffic and weather variables. With the arrival of the responders at the accident scene, more information is gained; therefore, the second stage model is activated, which updates the remaining accident duration time. We used eXtreme Gradient Boosting (XGBoost), a machine learning algorithm, and Shapley Additive exPlanations (SHAP) for making predictions and interpreting results, respectively. The results show that our framework predicts traffic levels with overall accuracies of around 80%, and duration models have high forecast accuracy with mean absolute percentage errors ranging between 7.26% and 21.59%. Overall, in the absence of accident information, SHAP values identified that weather factors, the traffic speed difference before and after an accident, traffic volume, and the percentage of heavy vehicles before the accident are the most important variables. However, accident variables, including the occurrence of injury or fatal accidents, rear-end collisions, and the number of involved vehicles, are among the most important variables in the second stage of the framework. The findings have practical implications for real-time traffic management of accident events. Road operators could manage post-accident traffic conditions more effectively, and road users could be alerted to take another route or manage their trip.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".