Bayesian Network Modeling for Causation Analysis of Traffic Accident

Fang Zong

2011-01-01

Abstract:A Bayesian network for traffic accident causation analysis was developed by structure and parameter learning,using correlation analysis,K2 algorithm and Bayesian method.Based on the Bayesian network,the interaction mechanism between the causing factors and the casualties of traffic accident was infered,and the effect of traffic control improvement on accident casualties reduction was analyzed.The results show that the Bayesian network can express the complicated relationship between the traffic accident and the causes,as well valuable information on how to take effective measures to reduce casualties of traffic accident.Moreover,the model has a high accuracy.The study can contribute to the development of traffic accident causality theory and the improvement of traffic safety situations.