Analysis Of Spatiotemporal Characteristics And Severity Modeling Of Taxi Traffic Violations

Traffic violations are one of the significant factors leading to road traffic accidents.Studying the patterns of drivers’ traffic violations and taking timely and effective measures can help reduce the occurrence of traffic violations in urban areas.Taxis are an essential component of urban road traffic.Compared to private car drivers,taxi drivers are more prone to traffic violations and poor driving habits,which can affect the overall traffic capacity of the city’s roads.The data used in this article is sourced from the electronic automatic capture devices for traffic violations in Nanchang City,covering the period from July 2020 to June 2021.These data include 43,622 records of traffic violations by taxis.Firstly,Arc GIS software is used to conduct spatiotemporal visualization analysis of taxi traffic violations in Nanchang City.Secondly,an analysis is conducted on the severity of traffic violations based on 11 influencing factors,including the time of violation,road conditions,environment,and taxi company attributes.A severity model for traffic violations is constructed,and the results of the model are interpreted using the Shapley Additive explanation(SHAP)method to explain the impact of each feature on the severity of traffic violations.Corresponding rectification recommendations are proposed based on the findings.(1)Spatial-temporal characteristics and hotspot analysis of taxi traffic violations in Nanchang City were conducted.Firstly,from a temporal perspective,it was found that taxi traffic violations primarily occur during the hour following peak congestion,with an increased proportion of severe violations during the early morning hours.Arc GIS’s kernel density analysis tool was used to visualize and analyze the six administrative districts separately,revealing that taxis are more prone to traffic violations at intersections on main roads,as well as near hospitals,schools,bus terminals,train stations,residential areas,and schools.Finally,the Moran I global spatial autocorrelation analysis method was employed to verify the spatial autocorrelation of taxi traffic violations in Nanchang City.Furthermore,a spatiotemporal evolution analysis of traffic violations was conducted.(2)A severity model for traffic violations was constructed.Firstly,the Balanced Bagging Classifier was used to address the data imbalance and reduce the imbalance ratio(IR)of the original data.Secondly,the grid search algorithm and ten-fold cross-validation were applied to optimize the hyperparameters of the decision tree C5.0,XG Boost,random forest,Ada Boost,and Multi-Layer Perceptron(MLP)models.The results showed that the random forest model outperformed the C5.0,XG Boost,Ada Boost,and MLP models in terms of various evaluation metrics.(3)Analyzing the results of the severity model for traffic violations using SHAP interpretation.The study found that functional districts,intersection of road segments,points of interest(POI),and road grade were significant factors influencing traffic violations,with SHAP values of 0.39,0.36,0.26,and 0.19,respectively.In order to reduce traffic violations by taxi drivers and improve road safety management,this research suggests that road traffic management authorities should inspect,correct,and improve signage and road markings.They should also ensure reasonable allocation of traffic light timings,check for obstructions of traffic lights by trees in old town areas,and adjust the height and angle of traffic signals.Additionally,developing more flexible and accurate parking spaces on side roads can be beneficial.Implementing these measures can help reduce traffic violations by taxis in Nanchang City and contribute to a more civilized transportation system.