A Study of Association Feature Mining for Risky Driving Behavior Considering the Influence of Traffic Signs

Risky driving behavior is an important factor in inducing chaotic road conditions and crash occurrences. The purpose of this study is to examine the association between risky driving behaviors and various factors, including roadway alignment, facility conditions, traffic operating conditions, and weather. First, the study uses kernel density analysis to explore the locations of risky driving behaviors. Next, the K-means algorithm to categorize continuous variables. Finally, the classical Apriori algorithm was used to mine the associative relationships of risky driving behaviors in different situations. It was found that frequent risky driving behaviors were mainly distributed near interchanges, which were related to the influence of merging areas, complex signs, and sign types. In addition, weather and road alignment were also found to be associated with risky driving behavior. The results of this study are useful for observing the association of risky driving behaviors in different situations, and then proposing targeted safety control.