A study on road accident prediction and contributing factors using explainable machine learning models: analysis and performance
Bibliographic record
Abstract
Road accidents are increasing worldwide and are causing millions of deaths each year. They impose significant financial and economic expenses on society. Existing research has mostly studied road accident prediction as a classification problem, which aims to predict whether a traffic accident may happen in the future or not without exploring the underneath relationships between the complicated factors contributing to road accidents. A number of research have been done to date to explore the importance of road accident contributing factors in relation to road accidents and their severity, however, only a few of those research have explored a subset of ensemble ML models and the New Zealand (NZ) road accident dataset. Therefore, in this paper, we have evaluated a set of machine learning (ML) models to predict road accident severity based on the most recent NZ road accident dataset. We have also analysed the predicted results and applied an explainable ML (XML) technique to evaluate the importance of road accident contributing factors. To predict road accidents with different injury severity, this work has considered different ensembles of ML models, like Random Forest (RF), Decision Jungle (DJ), Adaptive Boosting (AdaBoost), Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (L-GBM), and Categorical Boosting (CatBoost). New Zealand road accident data from 2016 through 2020 obtained from the New Zealand Ministry of Transport is used to perform this study. The comparison results show that RF is the best classifier with 81.45% accuracy, 81.68% precision, 81.42% recall, and 81.04% of F1-Score. Next, we have employed the Shapley value analysis as an XML technique to interpret the RF model performance at global and local levels. While the global level explanation provides the rank of the features’ contribution to severity classification, the local one is for exploring the use of features in the model. Furthermore, the Shapley Additive exPlanation (SHAP) dependence plot is used to investigate the relationship and interaction of the features towards the target variable prediction. Based on the findings, it can be said that the road category and number of vehicles involved in an accident significantly impact injury severity. The identified high-ranked features through SHAP analysis are used to retrain the ML models and measure their performance. The result shows 6%, 5%, and 8%, increase, respectively, in the performances of DJ, AdaBoost, and CatBoost models.
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 | 0.005 | 0.010 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.000 |
| 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".