應用個人違規紀錄預測交通事故發生之研究

• Main Authors: Chuang, Chih-Jen, 莊智仁
• Other Authors: Chang, Li-Yen
• Format: Others
• Language: zh-TW
• Published: 2003
• Online Access: http://ndltd.ncl.edu.tw/handle/03317922860882237496

碩士 === 國立嘉義大學 === 運輸與物流工程研究所 === 91 === Traffic accidents have been a serious problem for every country in the world. Each year more people were killed in the traffic accidents than any other accidents or natural disasters. According to the past statistics, more than 90% of traffic accidents were resulted from drivers’ errors or violations. Many studies have attempted to identify the contributing factors to traffic accidents in order to reduce traffic accidents. However, one of the important factors, traffic violation behavior, has not yet been carefully investigated. Therefore, the objectives of this study are to examine their relationships and also identify the traffic violations that significantly influence traffic accidents. This study collected the data of traffic accidents that occurred in Taipei during 2001 and 5-year traffic violation data for the involved drivers. 1:1 matched pairs logistic regression and artificial neural network analyses were applied to investigate the relationships between traffic accidents and traffic violations. The analysis results from the logistic regression models show many traffic violation behaviors, which can significantly increase the likelihood of having an accident for the drivers. These violation behaviors include illegally entering a one-way street, drunk driving, failing to grant right-of-way to vehicles or pedestrians, and vehicle equipment violations. The analysis results also demonstrate that artificial neural network, which is able to identify the relationship between traffic accidents and traffic violations, is an appropriate analysis tool in traffic accident analysis. The combined artificial neural network model, which uses statistically significant variables identified in the logistic regression model as the input neurons, has the better prediction capability in this particular application.