Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

An Explainable Machine Learning Approach to Traffic Accident Fatality Prediction

Md. Asif Khan Rifat, Ahmedul Kabir, Armana Sabiha Huq

TL;DR

This study presents a machine learning-based approach for classifying fatal and non-fatal road accident outcomes using data from the Dhaka metropolitan traffic crash database from 2017 to 2022, and demonstrates that LightGBM outperforms other models.

Abstract

Road traffic accidents (RTA) pose a significant public health threat worldwide, leading to considerable loss of life and economic burdens. This is particularly acute in developing countries like Bangladesh. Building reliable models to forecast crash outcomes is crucial for implementing effective preventive measures. To aid in developing targeted safety interventions, this study presents a machine learning-based approach for classifying fatal and non-fatal road accident outcomes using data from the Dhaka metropolitan traffic crash database from 2017 to 2022. Our framework utilizes a range of machine learning classification algorithms, comprising Logistic Regression, Support Vector Machines, Naive Bayes, Random Forest, Decision Tree, Gradient Boosting, LightGBM, and Artificial Neural Network. We prioritize model interpretability by employing the SHAP (SHapley Additive exPlanations) method, which elucidates the key factors influencing accident fatality. Our results demonstrate that LightGBM outperforms other models, achieving a ROC-AUC score of 0.72. The global, local, and feature dependency analyses are conducted to acquire deeper insights into the behavior of the model. SHAP analysis reveals that casualty class, time of accident, location, vehicle type, and road type play pivotal roles in determining fatality risk. These findings offer valuable insights for policymakers and road safety practitioners in developing countries, enabling the implementation of evidence-based strategies to reduce traffic crash fatalities.

An Explainable Machine Learning Approach to Traffic Accident Fatality Prediction

TL;DR

This study presents a machine learning-based approach for classifying fatal and non-fatal road accident outcomes using data from the Dhaka metropolitan traffic crash database from 2017 to 2022, and demonstrates that LightGBM outperforms other models.

Abstract

Road traffic accidents (RTA) pose a significant public health threat worldwide, leading to considerable loss of life and economic burdens. This is particularly acute in developing countries like Bangladesh. Building reliable models to forecast crash outcomes is crucial for implementing effective preventive measures. To aid in developing targeted safety interventions, this study presents a machine learning-based approach for classifying fatal and non-fatal road accident outcomes using data from the Dhaka metropolitan traffic crash database from 2017 to 2022. Our framework utilizes a range of machine learning classification algorithms, comprising Logistic Regression, Support Vector Machines, Naive Bayes, Random Forest, Decision Tree, Gradient Boosting, LightGBM, and Artificial Neural Network. We prioritize model interpretability by employing the SHAP (SHapley Additive exPlanations) method, which elucidates the key factors influencing accident fatality. Our results demonstrate that LightGBM outperforms other models, achieving a ROC-AUC score of 0.72. The global, local, and feature dependency analyses are conducted to acquire deeper insights into the behavior of the model. SHAP analysis reveals that casualty class, time of accident, location, vehicle type, and road type play pivotal roles in determining fatality risk. These findings offer valuable insights for policymakers and road safety practitioners in developing countries, enabling the implementation of evidence-based strategies to reduce traffic crash fatalities.
Paper Structure (15 sections, 2 equations, 6 figures, 2 tables)

This paper contains 15 sections, 2 equations, 6 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Literature Review
  3. Statistical and ML Modeling of Accident Outcomes
  4. Interpretable Models and Factors Analysis
  5. Methods
  6. Dataset Overview and Preprocessing
  7. Feature Selection
  8. Model Training
  9. Model Interpretation with SHAP
  10. Results and Discussion
  11. Model Evaluation
  12. Global Model Explanation
  13. Local Model Explanation
  14. Conclusion and Future Work
  15. Acknowledgement

Figures (6)

  • Figure 1: Framework of road accident fatality prediction with model interpretability using SHAP XAI.
  • Figure 2: Recursive feature elimination with SHAP (ShapRFECV) performance visualization.
  • Figure 3: ROC-AUC curve of different ML models for traffic accident fatality prediction.
  • Figure 4: (a) Beeswarm plot explaining LGBM on accident fatality; (b) Heatmap of SHAP values for all features across all samples in test set.
  • Figure 5: Dependence plots (a) impact of Date and Collision type on model output; (b) impact of Driver age and Vehicle type on model output.
  • ...and 1 more figures