Abstract:Although existing machine learning-based methods for traffic accident analysis can provide good quality results to downstream tasks, they lack interpretability which is crucial for this critical problem. This paper proposes an interpretable framework based on Bayesian Networks for traffic accident prediction. To enable the ease of interpretability, we design a dataset construction pipeline to feed the traffic data into the framework while retaining the essential traffic data information. With a concrete case study, our framework can derive a Bayesian Network from a dataset based on the causal relationships between weather and traffic events across the United States. Consequently, our framework enables the prediction of traffic accidents with competitive accuracy while examining how the probability of these events changes under different conditions, thus illustrating transparent relationships between traffic and weather events. Additionally, the visualization of the network simplifies the analysis of relationships between different variables, revealing the primary causes of traffic accidents and ultimately providing a valuable reference for reducing traffic accidents.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the lack of interpretability in traffic event (especially traffic accident) analysis. Although the existing machine - learning - based traffic event prediction methods can provide high - quality results for downstream tasks, they lack interpretability, which is crucial for such a critical issue. Specifically: 1. **Limitations of Existing Methods**: - Existing traffic event prediction methods (such as neural networks and random forests) can provide accurate prediction results, but lack transparency and interpretability. - Other powerful machine - learning algorithms (such as support vector machines) are also not easy to interpret. - Although deep - learning models perform well in simulating interconnected neurons, they are deficient in transparency and interpretability, which limits their application in reliable traffic analysis. 2. **Importance of the Problem**: - Traffic events (especially traffic accidents) have led to a large number of lives lost and huge economic losses. - The key to reducing traffic accidents lies in being able to predict the occurrence of accidents and understanding the factors that lead to accidents. - Interpretability is crucial for building trust, identifying errors, and improving reliability. 3. **Solutions Proposed in the Paper**: - The paper proposes an interpretable framework based on Bayesian Networks (BN) for traffic event prediction. - By designing a data set construction pipeline, traffic data is input into the framework while retaining necessary traffic information. - This framework can derive Bayesian Networks from the data set based on the causal relationship between weather and traffic events. - By visualizing the network structure, the analysis of the relationships between different variables is simplified, and the main causes of traffic accidents are revealed, thus providing a valuable reference for reducing traffic accidents. 4. **Specific Objectives**: - Provide competitive prediction accuracy. - Examine the probability changes of these events occurring under different conditions. - Demonstrate the transparent relationship between traffic and weather events. - Analyze the influence of a single factor on the probability of other events occurring. In summary, this paper aims to improve the interpretability of traffic event prediction by introducing the Bayesian network framework, so as to better understand the factors that lead to traffic accidents and provide a scientific basis for reducing traffic accidents.

Interpretable Traffic Event Analysis with Bayesian Networks

Bayesian Network for Red-Light-running Prediction at Signalized Intersections

A real-time explainable traffic collision inference framework based on probabilistic graph theory

Traffic accident severity prediction based on interpretable deep learning model

Data-Driven Bayesian Network Analysis of Railway Accident Risk

Real-Time Crash Risk Prediction using Long Short-Term Memory Recurrent Neural Network

Predicting Multiple Types of Traffic Accident Severity with Explanations: A Multi-Task Deep Learning Framework

Analyzing Traffic Crash Severity With Combination of Information Entropy and Bayesian Network

A probabilistic Bayesian inference model to investigate injury severity in automobile crashes

Learning Traffic Crashes as Language: Datasets, Benchmarks, and What-if Causal Analyses

Bayesian network modeling of accident investigation reports for aviation safety assessment

Using Bayesian Networks to Identify the Causal Effect of Speeding in Individual Vehicle/Pedestrian Collisions

Bayesian Network Methods For Traffic Flow Forecasting With Incomplete Data

Road Traffic Accident Risk Prediction and Key Factor Identification Framework Based on Explainable Deep Learning

Urban Traffic Accident Risk Prediction Revisited: Regionality, Proximity, Similarity and Sparsity

A Bayesian Network Approach To Time Series Forecasting Of Short-Term Traffic Flows

Traffexplainer: A Framework Towards GNN-based Interpretable Traffic Prediction

A novel feature extraction model for traffic injury severity and its application to Fatality Analysis Reporting System data analysis

Analysis of maritime transport accidents using Bayesian networks

Revisiting the Modifiable Areal Unit Problem in Deep Traffic Prediction with Visual Analytics

Explainable artificial intelligence in transport Logistics: Risk analysis for road accidents