ABSTRACT

With the rapid development of automobile industry in China, the number of automobile has grown up to 59.04 million till March 2008. Although the people's quality of life has been improved, the attendant traffic accidents become a major social hazard. In 2007 the number of deaths of traffic accidents reached 81,649. By studying on traffic accidents, the incentives of traffic accidents, and the reasons for the injured can be found.

For relatively late starting in China, the research on traffic accidents lacks detailed accident data and advanced research methods. In this paper The Traffic Accident Research System is designed and developed to save and share the data. The system can become a platform of accident data. Another studying is the use of data mining techniques to analyze the data. This method can enrich the functions of data research in system.

Firstly this paper analyzes four effect factors of traffic accident, persons, vehicles, roads and environment. Secondly this paper designed the data structure of system and realized it. After the input of data, this system has begun to have the functions of a data platform. In the end, a traffic accident data mining subsystem is designed. Two modules, the research of pedestrian protection based on multi-dimensional association rules and motorcycle accident prediction based on Bayesian network, are implemented.

Traffic Accident Information System, settling the lack of detailed accident data of current traffic accidents research, is highly practical.  Accidents data mining subsystem is a new attempt of data mining in the accidents study. For this, not only the expected research goal is achieved, but also a new approach for traffic accidents in-depth study is provided.

Key Word: traffic accident, data mining, Pedestrian Protection, multidimensional association rule, Bayesian network