The Palm distribution of traffic conditions and its application to accident risk assessment. (December 2016)

• Record Type: Journal Article
• Authors: Norros, Ilkka; Kuusela, Pirkko; Innamaa, Satu; Pilli-Sihvola, Eetu; Rajamäki, Riikka
• Is Part Of: Analytic methods in accident research. Volume 12(2016)
• Page Start: 48
• Page End: 65
• Publication Date: 2016-12
• Subjects: Palm distribution -- Traffic accident risk -- Traffic condition -- Weather condition -- Statistical testing; Accidents -- Research -- Methodology -- Periodicals; Accidents -- Prevention -- Periodicals; 363.100721
• DOI: 10.1016/j.amar.2016.10.002
• Languages: English
• ISSNs: 2213-6657

Abstract

Abstract: We introduce a method for assessing the influence of various road, weather and traffic conditions on traffic accidents. The idea is to contrast the distribution of conditions as seen by the driver involved in an accident with their distribution as seen by an arbitrary driver. The latter is considered as a variant of the notion of Palm probability of a point process, and it is easy to compute when road, weather and traffic measurement data are available. The method includes straightforward assessment of the statistical significance of the findings. We then study a single large example case, Ring-road I in Helsinki observed over five years, and present a comprehensive analysis of the influence of traffic, road and weather conditions on traffic accidents. Our results are in line with existing knowledge; for example, the traffic volume as such has hardly any influence on accidents, whereas the afternoon rush hours are considerably more risky than the morning ones, and heavy rain and snowfall as well as reduced visibility in general increase the accident risk substantially. The notion of Palm probability offers a transparent and uniform approach to such questions, and the proposed approach can be applied as a semi-automatic risk assessment tool prior to deeper analyses. Abstract : Highlights: The notion of an empirical Palm distribution of traffic conditions is introduced. The Palm distribution provides a new approach to accident risk assessment. The method is demonstrated on a 5-year dataset of Finland's most loaded road.