Study On Association Rules Mining Of Expressway Traffic Accidents Considering Data Heterogeneity

The number of road traffic accidents remains high,and the number of deaths from highway traffic accidents is as high as one million each year.Analyzing the factors affecting the occurrence of traffic accidents,exploring the internal causes of accidents,and taking targeted safety defense measures are the most direct and effective means to reduce the occurrence of traffic accidents and guarantee the level of traffic safety.Traffic accidents have many factors.After the accident,the traffic police department uses the accident information collection table to collect as many relevant factors as possible to provide data for traffic safety analysis.However,due to the existence of factors that have a significant impact on the accident but cannot be observed and recorded(such as the driver’s education level),there is a certain heterogeneity effect in the traffic accident data.It may be obtained without considering these factors in the accident analysis Inaccurate inference.In order to provide more accurate traffic safety decision support,this paper considers the heterogeneity of traffic accident data,and uses association rules algorithm to analyze the traffic accident data to mine the rules of accident occurrence.First,on the basis of summarizing the influencing factors of traffic accidents,this article defines the concept of accident data heterogeneity,analyzes the sources of data heterogeneity,compares the advantages and disadvantages of different models for processing data heterogeneity,and selects the potential category model as the treatment The method of data heterogeneity in this paper;Secondly,integrate the collected traffic accident data,extract 11 attribute variables that affect traffic accidents,perform descriptive statistics and coding,and establish a highway traffic accident data set;Next,through the potential category model,the accident data set is divided into three homogeneous accident groups,and the heterogeneity characteristics of the three accident groups are analyzed;Finally,the Apriori association rule algorithm is used to mine the association attribute rules of the three accident data groups and the overall accident set.The results show that: in the case of dividing three homogeneous accident groups,the potential category model can explain the driver’s unsafe driving behavior and the heterogeneity of the accident form better,and the effect of explaining the variables such as vehicle type,time and season is poor.In other words,the interpretation effect of variables such as age,gender,line shape,weather,road section and lane is poor;the Apriori algorithm is used to mine traffic accident data to obtain trailing collision,fixed object,side collision,scraping and other forms The regularity of traffic accidents and put forward targeted prevention suggestions.The research results of this paper are helpful for understanding the occurrence rules of traffic accidents,and can provide theoretical support for traffic accident safety improvement countermeasures.