Deep Learning Model for Crash Injury Severity Analysis Using Shapley Additive Explanation Values

Abstract

Analysis of traffic crash and associated data provides insights and assists with identification of cause-and-effect relationships with crash probabilities and outcomes. This study utilized eight years of police-reported Nebraska crash data using a deep neural network (DNN) to model crash injury severity outcomes. Prediction performances and model interpretability were examined. The developed DNN excelled in prediction accuracy, precision, and recall but was computationally intensive compared with a baseline multinomial logistic regression model. While the lack of interpretability power of deep learning models limits their usage, the adoption of SHapley Additive exPlanation (SHAP) values was an improvement. Conclusions drawn from the DNN model are generally consistent with the estimated baseline model. For instance, the variable total number of pedestrians was found significant in both scenarios of the multinomial logit model indicating a strong relationship with more severe crash injury outcomes. It was also found important in all three sets of parameters in DNN. SHAP values also allow in-depth analysis of prediction results on a single observation, such as the variable crash type (same direction sideswipe) contributing to classifying a single observation as property damage only. These findings are beneficial for making more informed transportation safety-related decisions.

Exploring the Determinants of Pedestrian Crash Severity: A Comparative Study Using Ordered Probit, Stacking Ensemble, and TabNet

“…ML and DL methods -Higher predictive accuracy: ML and DL methods can achieve superior predictive accuracy compared to statistical models, particularly when handling complex and non-linear relationships between variables or when dealing with large and complex datasets (Kang & Khattak, 2022;Komol et al, 2021).…”

“…-Computational resources: DL models typically demand more computational resources and longer training times in comparison to statistical and ML models (Kang & Khattak, 2022).…”

“…However, their study did not focus on pedestrian-related crashes, and they did not propose the TabNet interpretation results. Therefore, our study also contributes by introducing the interpretation of TabNet results using SHAP, a framework previously employed for interpreting DNN models in crash injury severity analysis by Kang et al (2022), for XGBoost models in similar studies by Chang et al (2022) and Li (2022) and by Islam and Abdel-Aty (2023) for long short-term memory (LSTM). By incorporating SHAP, we aim to provide deeper insights into the factors influencing pedestrian crash severity predictions using the TabNet model.…”

“…Machine learning (ML) models, including random forest, AdaBoost (Al-Mistarehi et al, 2022), XGBoost (Goswamy et al, 2023), decision trees (Effati & Vahedi Saheli, 2022), k-nearest neighbor, and ensemble models (L. Yang et al, 2022) have also been utilized. Additionally, deep learning (DL) models, like deep neural networks (DNN) (Kang & Khattak, 2022), have been explored for pedestrian crash severity analysis. To gain a better understanding of the application of these techniques, Table 1 presents an overview of the advantages and limitations of these methods used in pedestrian crash severity analysis.…”

Equity in Active and Shared Transportation: An Investigation of Barriers and Individual and Contextual Factors

“…SHAP improves the interpretability by estimating the positive or negative relationship for each factor with the dependent variable from the outcome of the model such as XGBoost. This technique has been used extensively in traffic safety studies, including crash injury prediction, to better interpret risk factors ( 45 , 46 ). In recent years, SHAP has also been used to estimate and interpret transportation mode preferences from a variety of data sources, including smart cards and travel surveys ( 47 , 48 ).…”