Research On Traffic Accident Causation And Accident Warning Based On Visual Analysis

As the number of vehicles worldwide increases year by year,traffic accidents are on the rise,which has become a problem of great concern to countries around the world.In order to improve traffic safety,China and developed countries in Europe and America have been strengthening traffic accident analysis and research.Not only does the exploration of accident causation and accident warning play a positive role in reducing the occurrence of accidents and damage caused by accidents,but also the laws related to accident causation can help relevant organizations to establish appropriate laws and regulations as well as develop theoretical frameworks and prevention plans related to accidents.The causes of traffic accidents include time,weather,geographic location,traffic flow,driver status,and whether there are multi-vehicle accidents,etc.These causes not only have multidimensional characteristics of multidimensional data,but also have strong dependency between attributes,so users cannot directly obtain the causes of accidents.Mining and early warning.For traffic accident causation mining and accident early warning,the main work done in this paper is as follows.1.Based on visual analytics,we perform feature analysis on accident data to extract spatiotemporal patterns,common features and uncertainties of accidents,and optimize the Apriori algorithm based on visual analytics and visually construct accident mining constraint sets to help users reduce analysis pressure and improve mining efficiency.2.Help users perceive the pattern of possible future accidents by constructing traffic accident warning models.Current accident prediction has not explained the impact of accident features on accident outcomes,while failing to provide users with a clearer understanding of the impact of accident features on accident severity.In this paper,we provide an early warning of accidents based on XGBOOST and SHAP and analyze the important features of accidents and how the features affect the severity of accidents.3.We designed and implemented a traffic accident causation mining and accident warning system to mine and analyze the patterns and laws of accident occurrence and help users to perceive the possible accident patterns in the future.Based on visual analysis technology,the system analyzes spatio-temporal patterns,common features and uncertainties of accidents,constructs a constraint set for accident mining visually,realizes accident causation mining and analysis,and realizes accident warning to warn users of possible dangers and explain the causes of accidents.Finally,case studies and user evaluations are used to demonstrate the practicality and innovation of this system,which is important for reducing accidents.