Study On Road Traffic Accident Influencing Factors Based On Association Rules And Spatial Autocorrelation

Because of the frequent traffic accidents,it has become one of the focuses of current traffic safety research to deeply analyze the influencing factors and reveal the influencing mechanism.However,the traditional analysis method is difficult to realize the potential correlation between multi-dimensional factors in the face of massive accident data.On the other hand,as the elements of traffic planning are gradually included in the scope of accident safety analysis,there are some factors analysis methods based on traffic zones.However,due to many uncertain factors,the accident has self-correlation in space.If this spatial correlation is ignored,the accuracy and robustness of accident safety analysis will be affected.Therefore,based on association rules and spatial autocorrelation,this paper studies the impact factors of accidents.The analysis of accident influencing factors based on association rules aims to explore the potential correlation law between influencing factors,and deeply analyze the cause of accidents,and provide reference for accident prevention.Tthe study based on spatial autocorrelation aims to consider the spatial autocorrelation of accidents,identify the key macro-planning factors affecting the frequency of accidents,and provide decision-making basis for traffic safety planning.The main tasks are as follows:(1)Based on mathematical statistics method to explore the distribution law and characteristics of traffic accidents,and preliminary analysis of the factors related to accidents.Firstly,the law of accident distribution is analyzed from the perspective of time and space.Secondly,the overall characteristics of the accident are analyzed from the perspective of accident level,shape and cause.Finally,based on the accident data,the factors affecting the accident are analyzed from four aspects:personnel,vehicle,road and environment.(2)Based on association rules to mine and analyze the relationship between the influencing factors of accidents.Firstly,the idea of mining association rules of accident influencing factors based on multidimensional data model is proposed.The star-shaped multidimensional data model suitable for mining association rules of accident influencing factors is constructed by using OLAP(Online Analytical Processing)technology.Then,based on the head table structure and FP-Tree pruning strategy,the FP-Growth algorithm is optimized,and the improved FP-Growth algorithm is used to mine and analyze the association rules between the whole factor of accident,single dimension and accident dimension.Finally,some suggestions for traffic accident prevention are put forward.(3)Modeling and analyzing the influencing factors of accidents based on spatial autocorrelation.Firstly,through the global Moran index test,it is found that there is significant autocorrelation in the spatial distribution of traffic accidents in the study area.Therefore,based on the negative binomial regression model,the Bayesian CAR(Conditional Autoregressive)residual term is introduced to construct an accident influencing factor analysis model considering spatial autocorrelation.Empirical results show that the density of main roads and intersections of limbs are the key factors leading to increased risk of traffic accidents in the study area.