Analyzing Winter Crash Dynamics Using Spatial Analysis and Crash Frequency Prediction Models with SHAP Interpretability
Bibliographic record
Abstract
This study investigates the application of machine learning (ML) to understand and mitigate winter road risks while addressing model interpretability. Using 26,970 winter crash records collected over four years in Edmonton, Canada, we developed and compared three ML-based winter crash frequency models: XGBoost, Random Forest, and LightGBM. To enhance interpretability, we applied SHapley Additive exPlanations (SHAP), providing insights into feature contributions. Our analysis incorporated micro-level variables such as collision records, weather conditions, and road characteristics, as well as macro-level variables such as land use patterns, spatial characteristics (via Hot Spot Analysis), and traffic exposure (estimated using Ordinary Kriging). Among the models tested, XGBoost outperformed others, achieving a testing R2 of 92.67%, MAE of 3.64, and RMSE of 5.77. SHAP analyses on XGBoost provided both global and local explanations. At a global level, road type, speed limit, and traffic enforcement cameras were identified as key factors influencing crash frequency while locally, distinct features of high- and low-crash locations were highlighted, supporting targeted risk mitigation strategies. By bridging the gap between model accuracy and interpretability, this study demonstrates the value of interpretable ML models in improving winter road safety, offering actionable insights for informed decision-making and resource allocation in winter road maintenance.
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 machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.012 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 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 source (direct Gemma or distilled Codex), 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".