Analysis of traffic accidents based on Spark and causal inference

Abstract

Traffic accidents have occurred frequently in recent years, causing great losses to personal and property safety. Studying traffic accident data is helpful to identify key factors of traffic accidents from big data. This paper processes and calculates big data based on the Spark platform. By introducing causal inference into the analysis of traffic accidents, it establishes the causal relationships between 17 factors and the severity of traffic accidents, thereby analyzing the root causes and intermediate causes of the accidents. In addition, this paper also conducts an intervention study to evaluate the influence weight of each factor. The study finds that the physical conditions of pedestrians and weather conditions are inferred to be the root causes, and the others are intermediate causes. Besides that, the presence of police force and reduced traffic volume are considered to be the best ways to reduce traffic accidents. Therefore, this article believes that in real life, we should reduce the incidence of traffic accidents by controlling traffic flow and increasing patrol or on-duty police force. These findings provide a scientific basis for traffic management departments to develop more effective traffic safety strategies.