Influencing Factors Analysis And Prediction Of The Severity Of Taxi Drivers' Traffic Accidents

Driven by the rapid economic development in China,the vehicle ownership has also been increasing rapidly.As a result,the number of traffic accidents in China has become more frequent.According to the statistics of the National Bureau of Statistics,216,585 automobile accidents occurred in 2016,and resulted in 58,803 fatalities.According to the research findings of the Indian University of America's impact factors,92% of accidents are related to human factors.A British survey showed that 95% of accidents are related to drivers.The research data in China shows that the driver accounted for 87.5% of the total number of accidents.Therefore,the driver factor is undoubtedly the most critical factor among the many factors affecting traffic accidents.In addition,taxi drivers as a broad group,there are great differences in traffic flow characteristics,driving purposes and driving psychology compared to ordinary private car drivers.It is obviously unreasonable to analyze taxi drivers mixed with ordinary private car drivers in many traffic analyses.Based on the above background,this article mainly takes the taxi driver as the research object,carries on the accident analysis and the forecast.From 2014 to 2015,we conducted a taxi accident survey in Xi'an,Xining,Changchun,and Shantou cities,and conducted exploratory factor analysis on collected taxi driver accident data(which collected 1010 valid data),and a structural equation model of traffic accidents was constructed to explore the main influencing factors of taxi accident severity and the influence of each factor on the severity of the accident.Aiming at the main accident influencing factors analyzed for the structural equation model,using the accident data of the training set,we use the generalized and ordered logit model(gologit)in statistics and the random forest in the machine learning to establish a taxi accident severity prediction model,respectively.And using the goodness of fit and other methods to test the model.Then,the taxi accident severity prediction model is used to predict the taxi driver data in the test and evaluate the model forecast results.Finally,the two models were compared and analyzed.It was found that random forests have better balance and are more suitable for accident prediction.By constructing a model for predicting the severity of accidents,it is possible to assess the potential risk of driving for rental drivers.Based on the analysis of the causes of taxi accidents based on the structural equation model,the 14 main influencing factors of taxi accidents are summarized and classified into three categories: personal attribute factors,driving behavior factors and working status factors.Taxi accident prevention measures and suggestions are proposed separately.