Analysis And Application Of Taxi Driver's Traffic Accident Based On Data Mining

Taxis have the characteristics of flexibility,high mobility and convenience in urban traffic.Taxi share in Beijing,Shanghai,Shenzhen and other big cities accounts for 25%-30%,which shows that taxis play an important role in modern urban life.As far as taxi traffic accidents are concerned,human factors account for a large proportion of accidents,among which the driver's factors are particularly important to traffic safety,while the accident rate caused by objective factors such as traffic environment or road geometry is relatively low.Therefore,from the perspective of taxi drivers,it is of great significance for traffic safety to analyze the factors affecting accident rate.In order to reduce the rate of taxi accidents,this paper takes taxi drivers as the research object,and constructs a risk level prediction model for taxi drivers.According to the taxi driver data obtained from the survey,the correlation analysis and exploratory factor analysis of various factors related to the driver are carried out,including the analysis of the internal relationship between the severity of taxi driver accidents and the influencing factors,and the important factors affecting the severity of taxi driver traffic accidents are explored.On this basis,data mining technology is used.This paper constructs a risk assessment model for taxi drivers and predicts the driver's risk level,including four single algorithm models: Logistic sequential regression,decision tree,support vector machine and BP neural network,and three integrated algorithm models of Bagging,Boosting and random forest.For the above models,through the comprehensive comparative analysis of accuracy,kappa value,stability and complexity,it is found that the BP neural network model performs best in the above seven models.In order to further optimize the BP neural network model in the single-algorithm model,the particle swarm optimization algorithm is used to improve the prediction accuracy of the model.It is proved that the prediction accuracy of the model is not only improved,but also more stable.Finally,combined with the practical application background of taxi driver's traffic accident risk in China,according to the risk prediction model of taxi driver constructed by particle swarm optimization BP neural network,effective measures to prevent taxi driver's traffic accident are put forward,in order to reduce the incidence of taxi accident.