文章基本信息

标题：Association knowledge for fatal run-off-road crashes by Multiple Correspondence Analysis
本地全文：下载
作者：Subasish Das ; Xiaoduan Sun
期刊名称：IATSS Research
印刷版ISSN：0386-1112
出版年度：2016
卷号：39
期号：2
页码：146-155
DOI：10.1016/j.iatssr.2015.07.001
出版社：Elsevier
摘要：In 2013, 346 out of 616 fatal crashes in Louisiana were single vehicle crashes with Run-Off-Road (ROR) crashes being the most common type of single vehicle crash. In order to create effective countermeasures for reducing the number of fatal single vehicle ROR crashes, it is important to identify any associated key factors that can quantitatively assess the performance of roads, vehicles and humans. This research uses Multiple Correspondence Analysis (MCA), a multidimensional descriptive data analysis method that associates a combination of factors based on their relative distance in a two dimensional plane, to analyze eight years (2004-2011) of fatal ROR crashes in Louisiana. This method measures important contributing factors and their degree of association. The results revealed that drivers of lightweight trucks, drivers on undivided state highways, male drivers in passenger-vehicles at dawn, older female (65-74) drivers in non-passenger vehicles, older drivers facing hardship to yield in partial access control zones, and drivers with poor reaction time due to impaired driving were closely associated with fatal ROR crashes. Results of the MCA method can help researchers select the most effective crash countermeasures. Further work on the degree of association between the identified crash contributing factors can help safety management systems develop the most efficient crash reduction strategies.
关键词：ROR crashes; Multiple Correspondence Analysis; Dimensionality reduction; Cloud of combination groups