Time series analysis in road safety research uisng state space methods
BIJLEVELD
Type de document
THESE
Langue
anglais
Auteur
BIJLEVELD
Résumé / Abstract
In this thesis we present a comprehensive study into novel time series models for aggregated road safety data. The models are mainly intended for analysis of indicators relevant to road safety, with a particular focus on how to measure these factors. Such developments may need to be related to or explained by external influences. It is also possible to make forecasts using the models. Relevant indicators include the number of persons killed permonth or year. These statistics are closely watched by government agencies and the public, and their relevance to society is not disputed. A large body of research is devoted to the improvement of road safety. To that end, changes in the number of accidents or victims are often attempted to be explained by (changes in) factors such as exposure, policy, driving under the influence of alcohol, speeding by drivers. Some factors such as policy changes can be directly observed (although compliance with policy and law may not). Other factors can be observed in theory but in practice their measurement is either difficult or very expensive. Examples of such factors are exposure, which ismeasured using surveys and vehicle counting systems, and percentage of drivers exceeding the legal blood alcohol concentration limit, which is measured using road side surveys. Finally, some factors are even harder to observe such as driver skill or experience.
Editeur
SWOV
